Up next in 10
Many oil and gas operating businesses have used digital oilfield apps to automate workflows and expedite procedures to maximize oil and gas output in real time. Most of these applications are installed utilizing conventional on-premises systems, where scalability, accessibility, and maintenance provide significant obstacles to effective operation. The industry still has issues with data integration from many data sources, releasing consolidated data for consumption, and orchestrating cross-domain workflows for that data.
Digital transformation strategies have created a path for cloud-based solutions, which have gained popularity recently in the oil and gas sector due to their increased accessibility, ease of customization, higher system stability, and scalability to accommodate greater volumes of data. An operational production data foundation system that aggregates production and equipment data from several organizational departments is necessary to tackle the challenges mentioned earlier.
This talk will focus on how operational production data foundation, hosted on the cloud, offers the underlying infrastructure, services, and interfaces needed to support and unify workflow orchestration and production data ingestion. Additionally, by bringing together digital concepts and common domains, it will enhance collaboration between individuals in different roles, including production engineers, software developers, data scientists, drilling engineers, and reservoir engineers, who work with production operations products and solutions. The main focus is going to be sharing knowledge of patented work that helped to solve data ingestion, canonical production models using bi-temporality and orchestration engine.
🔗 Conference Website: https://softwarearchitecture.live
📺 CSharp TV - Dev Streaming Destination http://csharp.tv
🌎 C# Corner - Community of Software and Data Developers
https://www.c-sharpcorner.com
#CSharpTV #CSharpCorner #CSharp #SoftwareArchitectureConf
Show More Show Less View Video Transcript
0:03
okay all right so uh I will start with
0:06
okay all right so uh I will start with
0:06
okay all right so uh I will start with my brief introduction first so my name
0:08
my brief introduction first so my name
0:08
my brief introduction first so my name is Abaya I'm working as a software team
0:11
is Abaya I'm working as a software team
0:11
is Abaya I'm working as a software team leader with ISJ short name is SLB and
0:16
leader with ISJ short name is SLB and
0:16
leader with ISJ short name is SLB and today's topic is U scaling oil field
0:19
today's topic is U scaling oil field
0:19
today's topic is U scaling oil field production operation using cloud
0:21
production operation using cloud
0:21
production operation using cloud computing so I completely agree with the
0:25
computing so I completely agree with the
0:25
computing so I completely agree with the last presenter and he focuses
0:28
last presenter and he focuses
0:28
last presenter and he focuses on uh the requirement and keep it simple
0:31
on uh the requirement and keep it simple
0:31
on uh the requirement and keep it simple so yeah that is also I follow in
0:33
so yeah that is also I follow in
0:33
so yeah that is also I follow in day-to-day life I mean I usually build
0:37
day-to-day life I mean I usually build
0:37
day-to-day life I mean I usually build Solutions incrementally not everything
0:39
Solutions incrementally not everything
0:39
Solutions incrementally not everything on single day I mean I start with
0:42
on single day I mean I start with
0:42
on single day I mean I start with something small then keep it rating over
0:44
something small then keep it rating over
0:44
something small then keep it rating over it so yeah I completely agree with
0:46
it so yeah I completely agree with
0:46
it so yeah I completely agree with whatever he he presented okay so before
0:51
whatever he he presented okay so before
0:51
whatever he he presented okay so before going into the main topic I would like
0:54
going into the main topic I would like
0:54
going into the main topic I would like to give a brief introduction about oil
0:57
to give a brief introduction about oil
0:57
to give a brief introduction about oil and gas business so for the folks who
1:00
and gas business so for the folks who
1:00
and gas business so for the folks who don't know s l is world's number one
1:03
don't know s l is world's number one
1:03
don't know s l is world's number one company in Upstream oil and gas services
1:07
company in Upstream oil and gas services
1:07
company in Upstream oil and gas services so around the globe there are mainly two
1:10
so around the globe there are mainly two
1:11
so around the globe there are mainly two kind of companies dealing with oil and
1:13
kind of companies dealing with oil and
1:13
kind of companies dealing with oil and gas business the first part of com
1:16
gas business the first part of com
1:16
gas business the first part of com company are basically oil and gas
1:18
company are basically oil and gas
1:18
company are basically oil and gas company who basically owns the oil well
1:21
company who basically owns the oil well
1:21
company who basically owns the oil well and these oil well could be anywhere on
1:24
and these oil well could be anywhere on
1:24
and these oil well could be anywhere on Earth or in offshore in middle of the
1:26
Earth or in offshore in middle of the
1:26
Earth or in offshore in middle of the sea and second of the company which are
1:29
sea and second of the company which are
1:30
sea and second of the company which are basically operating in these oil and gas
1:32
basically operating in these oil and gas
1:32
basically operating in these oil and gas well so I'm from the second part I'm
1:35
well so I'm from the second part I'm
1:35
well so I'm from the second part I'm working for the service provider so L is
1:38
working for the service provider so L is
1:38
working for the service provider so L is a world's number one Service Pro
1:40
a world's number one Service Pro
1:40
a world's number one Service Pro provider and what we do we basically
1:42
provider and what we do we basically
1:42
provider and what we do we basically deals with uh up Stream So if you want
1:45
deals with uh up Stream So if you want
1:45
deals with uh up Stream So if you want to know what is up stream so I'm showing
1:48
to know what is up stream so I'm showing
1:48
to know what is up stream so I'm showing on the screen so the left hand side that
1:51
on the screen so the left hand side that
1:51
on the screen so the left hand side that call up scam the middle says mid stream
1:54
call up scam the middle says mid stream
1:54
call up scam the middle says mid stream and the right one say down stream so
1:57
and the right one say down stream so
1:57
and the right one say down stream so what we do basically so we basically
2:00
what we do basically so we basically
2:00
what we do basically so we basically involved in everything whatever involves
2:02
involved in everything whatever involves
2:02
involved in everything whatever involves in drilling the oil well and extracting
2:05
in drilling the oil well and extracting
2:05
in drilling the oil well and extracting the oil from under the Earth and this
2:09
the oil from under the Earth and this
2:09
the oil from under the Earth and this oil well as I said it could be a middle
2:11
oil well as I said it could be a middle
2:11
oil well as I said it could be a middle of the sea it could be on some desert or
2:13
of the sea it could be on some desert or
2:13
of the sea it could be on some desert or some remote area so everything involved
2:17
some remote area so everything involved
2:17
some remote area so everything involved including some tools and Technologies
2:19
including some tools and Technologies
2:19
including some tools and Technologies and hardware and software everything so
2:22
and hardware and software everything so
2:22
and hardware and software everything so we provide all kind of Technologies and
2:25
we provide all kind of Technologies and
2:25
we provide all kind of Technologies and when I'm saying skill set you will find
2:28
when I'm saying skill set you will find
2:28
when I'm saying skill set you will find petroleum engineers geophysicist or
2:31
petroleum engineers geophysicist or
2:31
petroleum engineers geophysicist or petrophysicist geologist and software
2:34
petrophysicist geologist and software
2:34
petrophysicist geologist and software Engineers data scientist machine
2:35
Engineers data scientist machine
2:35
Engineers data scientist machine learning Engineers so everyone working
2:37
learning Engineers so everyone working
2:37
learning Engineers so everyone working as a team and our single goal is to
2:40
as a team and our single goal is to
2:40
as a team and our single goal is to optimize the oil production so we want
2:43
optimize the oil production so we want
2:43
optimize the oil production so we want to apply all
2:44
to apply all
2:44
to apply all Technologies uh together so that we can
2:47
Technologies uh together so that we can
2:47
Technologies uh together so that we can optimize the oil
2:49
optimize the oil
2:49
optimize the oil production and uh as you can see on left
2:52
production and uh as you can see on left
2:52
production and uh as you can see on left hand side I'm showing two two pictures
2:54
hand side I'm showing two two pictures
2:54
hand side I'm showing two two pictures one is the onshore pump check you will
2:56
one is the onshore pump check you will
2:56
one is the onshore pump check you will mostly see that Rod is basically going
2:59
mostly see that Rod is basically going
3:00
mostly see that Rod is basically going up and down it is basically drilling the
3:02
up and down it is basically drilling the
3:02
up and down it is basically drilling the oil oil well and the other picture says
3:04
oil oil well and the other picture says
3:04
oil oil well and the other picture says okay someone using some helicopter and
3:07
okay someone using some helicopter and
3:08
okay someone using some helicopter and landing on that uh pad location where
3:11
landing on that uh pad location where
3:11
landing on that uh pad location where there is some uh uh instruments are
3:14
there is some uh uh instruments are
3:14
there is some uh uh instruments are installed in middle of the sea and they
3:15
installed in middle of the sea and they
3:16
installed in middle of the sea and they are basically producing the oil and
3:19
are basically producing the oil and
3:19
are basically producing the oil and there then there are Midstream companies
3:21
there then there are Midstream companies
3:21
there then there are Midstream companies these companies are basically
3:23
these companies are basically
3:23
these companies are basically responsible for processing and storage I
3:26
responsible for processing and storage I
3:26
responsible for processing and storage I mean once the oil is produced then
3:28
mean once the oil is produced then
3:28
mean once the oil is produced then someone should be there to to transfer
3:30
someone should be there to to transfer
3:30
someone should be there to to transfer it so these companies are mostly uh
3:33
it so these companies are mostly uh
3:33
it so these companies are mostly uh build the pipelines and transport and
3:36
build the pipelines and transport and
3:36
build the pipelines and transport and trucks and the Third Kind of companies
3:38
trucks and the Third Kind of companies
3:38
trucks and the Third Kind of companies are Downstream companies so once uh this
3:42
are Downstream companies so once uh this
3:42
are Downstream companies so once uh this oil is already uh reached to the
3:45
oil is already uh reached to the
3:45
oil is already uh reached to the distribution C Center so these companies
3:47
distribution C Center so these companies
3:47
distribution C Center so these companies are basically refining the oil and after
3:50
are basically refining the oil and after
3:50
are basically refining the oil and after that they also deal with the sales and
3:52
that they also deal with the sales and
3:52
that they also deal with the sales and marketing so today's topic is mostly on
3:56
marketing so today's topic is mostly on
3:56
marketing so today's topic is mostly on upstream and about uh about a
4:01
upstream and about uh about a
4:01
upstream and about uh about a cloud-based solution what we built and
4:04
cloud-based solution what we built and
4:04
cloud-based solution what we built and it basically helped to optimize oil
4:07
it basically helped to optimize oil
4:07
it basically helped to optimize oil production so I would like to start with
4:11
production so I would like to start with
4:11
production so I would like to start with some terms like what is a digital oil
4:13
some terms like what is a digital oil
4:13
some terms like what is a digital oil oil field so digital oil oil field is
4:15
oil field so digital oil oil field is
4:15
oil field so digital oil oil field is just a fancy name what we use I mean
4:18
just a fancy name what we use I mean
4:18
just a fancy name what we use I mean it's nothing new it's there in the
4:20
it's nothing new it's there in the
4:20
it's nothing new it's there in the industry since last more than 25 years
4:23
industry since last more than 25 years
4:23
industry since last more than 25 years but uh since last more than 10 years
4:25
but uh since last more than 10 years
4:25
but uh since last more than 10 years everyone started talking about a cloud
4:28
everyone started talking about a cloud
4:28
everyone started talking about a cloud and now we haveck at GPT and machine
4:30
and now we haveck at GPT and machine
4:30
and now we haveck at GPT and machine learning and data scientist so digital
4:33
learning and data scientist so digital
4:33
learning and data scientist so digital term I mean people started correlating
4:36
term I mean people started correlating
4:36
term I mean people started correlating it with a cloud and machine learning and
4:38
it with a cloud and machine learning and
4:38
it with a cloud and machine learning and everything but for us the digital oil
4:40
everything but for us the digital oil
4:40
everything but for us the digital oil field is anything related to computer so
4:43
field is anything related to computer so
4:43
field is anything related to computer so digital oil field I mean even before the
4:45
digital oil field I mean even before the
4:45
digital oil field I mean even before the cloud cloud com Computing it was there
4:49
cloud cloud com Computing it was there
4:49
cloud cloud com Computing it was there but now we are using it more proactively
4:53
but now we are using it more proactively
4:53
but now we are using it more proactively and it is related to using any digital
4:56
and it is related to using any digital
4:56
and it is related to using any digital Technologies it could be software or
4:58
Technologies it could be software or
4:58
Technologies it could be software or Hardware or data analytics and the end
5:01
Hardware or data analytics and the end
5:01
Hardware or data analytics and the end goal is to enhance the exploration
5:03
goal is to enhance the exploration
5:03
goal is to enhance the exploration production and management of these oil
5:06
production and management of these oil
5:06
production and management of these oil gas field and uh the general operation
5:09
gas field and uh the general operation
5:09
gas field and uh the general operation involved I mean if we are collecting a
5:12
involved I mean if we are collecting a
5:12
involved I mean if we are collecting a lot of data from the sensors installed
5:14
lot of data from the sensors installed
5:14
lot of data from the sensors installed in this oil oil field so basically
5:17
in this oil oil field so basically
5:17
in this oil oil field so basically artificial intelligence is involved and
5:19
artificial intelligence is involved and
5:19
artificial intelligence is involved and internet of things are also involved so
5:22
internet of things are also involved so
5:22
internet of things are also involved so as you can see the right hand picture I
5:24
as you can see the right hand picture I
5:24
as you can see the right hand picture I generated from AI so this picture is
5:28
generated from AI so this picture is
5:28
generated from AI so this picture is basically showing a typical oil oil
5:31
basically showing a typical oil oil
5:31
basically showing a typical oil oil field where several peoples are working
5:33
field where several peoples are working
5:33
field where several peoples are working and you can also see some Hardwares are
5:36
and you can also see some Hardwares are
5:36
and you can also see some Hardwares are installed some computer schemes are
5:37
installed some computer schemes are
5:37
installed some computer schemes are installed they connect it to some some
5:40
installed they connect it to some some
5:40
installed they connect it to some some satellite so that they can get the
5:41
satellite so that they can get the
5:41
satellite so that they can get the internet connection and everything and
5:43
internet connection and everything and
5:43
internet connection and everything and they're always sending some data to the
5:45
they're always sending some data to the
5:45
they're always sending some data to the remote oil field operation so today's
5:48
remote oil field operation so today's
5:48
remote oil field operation so today's topic is about the similar application
5:50
topic is about the similar application
5:50
topic is about the similar application which is installed and which is
5:53
which is installed and which is
5:53
which is installed and which is basically uh optimizing this digital oil
5:55
basically uh optimizing this digital oil
5:55
basically uh optimizing this digital oil feed
5:57
feed
5:57
feed operation so I would like to start with
6:00
operation so I would like to start with
6:00
operation so I would like to start with the challenges the requirement what we
6:02
the challenges the requirement what we
6:02
the challenges the requirement what we learned in the last session as well so
6:05
learned in the last session as well so
6:05
learned in the last session as well so the challenges and requirement for
6:07
the challenges and requirement for
6:07
the challenges and requirement for production operation means uh because
6:09
production operation means uh because
6:09
production operation means uh because this operation is highly complex it
6:11
this operation is highly complex it
6:11
this operation is highly complex it basically involves a lot many people
6:13
basically involves a lot many people
6:13
basically involves a lot many people from different business roles and from
6:16
from different business roles and from
6:16
from different business roles and from different geographies and different
6:18
different geographies and different
6:18
different geographies and different workflows are always applied so these
6:21
workflows are always applied so these
6:21
workflows are always applied so these problems are typically solved using a
6:23
problems are typically solved using a
6:23
problems are typically solved using a diverse or disconnected set of software
6:25
diverse or disconnected set of software
6:25
diverse or disconnected set of software application and tools and uh before
6:28
application and tools and uh before
6:28
application and tools and uh before introducing the cloud computing part we
6:30
introducing the cloud computing part we
6:31
introducing the cloud computing part we were still doing this operation but at
6:33
were still doing this operation but at
6:33
were still doing this operation but at less scale and uh so the main challenge
6:38
less scale and uh so the main challenge
6:38
less scale and uh so the main challenge for with the application the existing
6:40
for with the application the existing
6:40
for with the application the existing application was I mean uh the collection
6:43
application was I mean uh the collection
6:43
application was I mean uh the collection of data was not easy and to make this
6:46
of data was not easy and to make this
6:46
of data was not easy and to make this data available for running a scalable
6:48
data available for running a scalable
6:48
data available for running a scalable workflow that was not a straightforward
6:50
workflow that was not a straightforward
6:50
workflow that was not a straightforward because our data was scattered and there
6:52
because our data was scattered and there
6:52
because our data was scattered and there was so many on-prem uh desktop based
6:56
was so many on-prem uh desktop based
6:56
was so many on-prem uh desktop based application was was running and
6:58
application was was running and
6:58
application was was running and collecting so it was not very much
7:00
collecting so it was not very much
7:00
collecting so it was not very much scalable second thing was in oil oil
7:03
scalable second thing was in oil oil
7:03
scalable second thing was in oil oil field the data frequency could be very
7:06
field the data frequency could be very
7:06
field the data frequency could be very different few data could be a second
7:09
different few data could be a second
7:09
different few data could be a second best data or maybe less than second
7:11
best data or maybe less than second
7:11
best data or maybe less than second based data few data could be monthly
7:15
based data few data could be monthly
7:15
based data few data could be monthly yearly weekly or daily and there are
7:18
yearly weekly or daily and there are
7:18
yearly weekly or daily and there are several conventions followed by
7:20
several conventions followed by
7:20
several conventions followed by different oil companies around the globe
7:22
different oil companies around the globe
7:22
different oil companies around the globe so whenever we we we have some uh
7:26
so whenever we we we have some uh
7:26
so whenever we we we have some uh existing desktop based application then
7:29
existing desktop based application then
7:29
existing desktop based application then each application talks their own data
7:31
each application talks their own data
7:31
each application talks their own data standard so whenever we are transferring
7:34
standard so whenever we are transferring
7:34
standard so whenever we are transferring data from one app to other we always
7:36
data from one app to other we always
7:36
data from one app to other we always need to build some some kind of
7:38
need to build some some kind of
7:38
need to build some some kind of connectors or or adapters so that one
7:41
connectors or or adapters so that one
7:41
connectors or or adapters so that one application understand what the other
7:43
application understand what the other
7:43
application understand what the other application is talking about and it was
7:46
application is talking about and it was
7:46
application is talking about and it was uh very hard to maintain it was not very
7:49
uh very hard to maintain it was not very
7:49
uh very hard to maintain it was not very much easily accessible and poor
7:52
much easily accessible and poor
7:52
much easily accessible and poor scalability and it was leading to
7:55
scalability and it was leading to
7:55
scalability and it was leading to non-productive time and data quality
7:57
non-productive time and data quality
7:57
non-productive time and data quality issue as well and at the end it was uh
8:00
issue as well and at the end it was uh
8:00
issue as well and at the end it was uh giving some inconsistencies as well
8:02
giving some inconsistencies as well
8:02
giving some inconsistencies as well because in this on on on Prime app I
8:05
because in this on on on Prime app I
8:05
because in this on on on Prime app I mean if there's a data became in
8:08
mean if there's a data became in
8:08
mean if there's a data became in inconsistent then I need to import that
8:11
inconsistent then I need to import that
8:11
inconsistent then I need to import that data again and connect it to the several
8:13
data again and connect it to the several
8:13
data again and connect it to the several application a couple of times so that
8:15
application a couple of times so that
8:15
application a couple of times so that was not a a good way to run these work
8:18
was not a a good way to run these work
8:18
was not a a good way to run these work workflows and other common complaint
8:22
workflows and other common complaint
8:22
workflows and other common complaint from customer was they always see a
8:23
from customer was they always see a
8:23
from customer was they always see a missing data or incomplete data so they
8:27
missing data or incomplete data so they
8:27
missing data or incomplete data so they they were missing a complete and
8:28
they were missing a complete and
8:28
they were missing a complete and consistent View of data always so these
8:32
consistent View of data always so these
8:32
consistent View of data always so these are the challenge so okay I talked
8:35
are the challenge so okay I talked
8:35
are the challenge so okay I talked enough about data and data about
8:36
enough about data and data about
8:36
enough about data and data about frequencies at all but I would like to
8:39
frequencies at all but I would like to
8:39
frequencies at all but I would like to give some context how how does that data
8:43
give some context how how does that data
8:43
give some context how how does that data look look like so on very high level I
8:45
look look like so on very high level I
8:45
look look like so on very high level I would like to split the data into the
8:48
would like to split the data into the
8:48
would like to split the data into the two in the two uh uh two kind of data I
8:53
two in the two uh uh two kind of data I
8:53
two in the two uh uh two kind of data I one is the time series data other one is
8:55
one is the time series data other one is
8:55
one is the time series data other one is the structural data when I'm saying time
8:57
the structural data when I'm saying time
8:57
the structural data when I'm saying time series data so when whenever we are
9:00
series data so when whenever we are
9:00
series data so when whenever we are drilling the oil well and we are putting
9:02
drilling the oil well and we are putting
9:02
drilling the oil well and we are putting a lot many tools inside the earth so
9:05
a lot many tools inside the earth so
9:05
a lot many tools inside the earth so with that drilling pipeline there are
9:07
with that drilling pipeline there are
9:07
with that drilling pipeline there are lot many sensors installed and these
9:09
lot many sensors installed and these
9:09
lot many sensors installed and these sensors are basically capturing a lot of
9:14
sensors are basically capturing a lot of
9:14
sensors are basically capturing a lot of data points from beneath the Earth these
9:17
data points from beneath the Earth these
9:17
data points from beneath the Earth these data points could be in simple terms of
9:19
data points could be in simple terms of
9:19
data points could be in simple terms of pressure or temperature or the gas flow
9:22
pressure or temperature or the gas flow
9:23
pressure or temperature or the gas flow rate or oil flow rate or water flow rate
9:25
rate or oil flow rate or water flow rate
9:25
rate or oil flow rate or water flow rate and typically this time space data looks
9:28
and typically this time space data looks
9:28
and typically this time space data looks like what I'm sh going on the right hand
9:31
like what I'm sh going on the right hand
9:31
like what I'm sh going on the right hand side the first one is say okay there is
9:33
side the first one is say okay there is
9:33
side the first one is say okay there is acquisition time which is which is
9:36
acquisition time which is which is
9:36
acquisition time which is which is associated with every data point let's
9:38
associated with every data point let's
9:38
associated with every data point let's say I'm showing that uh uh some value
9:41
say I'm showing that uh uh some value
9:42
say I'm showing that uh uh some value let's say some uh oil production volume
9:44
let's say some uh oil production volume
9:44
let's say some uh oil production volume coming so I'm showing 2.3 I assigning
9:47
coming so I'm showing 2.3 I assigning
9:47
coming so I'm showing 2.3 I assigning the unit of measurement as well and I'm
9:50
the unit of measurement as well and I'm
9:50
the unit of measurement as well and I'm also assigning the kind of data is it
9:52
also assigning the kind of data is it
9:52
also assigning the kind of data is it double or string so that my storage
9:54
double or string so that my storage
9:54
double or string so that my storage system or my workflows also understand
9:56
system or my workflows also understand
9:56
system or my workflows also understand it I'm also assigning The Source
9:58
it I'm also assigning The Source
9:58
it I'm also assigning The Source acquisition time I mean okay uh Source
10:00
acquisition time I mean okay uh Source
10:01
acquisition time I mean okay uh Source acquisition time means when I really see
10:03
acquisition time means when I really see
10:03
acquisition time means when I really see the data on my cloud so one agent
10:06
the data on my cloud so one agent
10:06
the data on my cloud so one agent acquisition time means the acquisition
10:08
acquisition time means the acquisition
10:08
acquisition time means the acquisition time when the onframe Tool which is
10:11
time when the onframe Tool which is
10:12
time when the onframe Tool which is basically streaming that data point and
10:14
basically streaming that data point and
10:15
basically streaming that data point and the source acquisition when I start
10:16
the source acquisition when I start
10:16
the source acquisition when I start seeing that data point on cloud so there
10:19
seeing that data point on cloud so there
10:19
seeing that data point on cloud so there are typical key characteristic for this
10:21
are typical key characteristic for this
10:21
are typical key characteristic for this time series data which is important for
10:23
time series data which is important for
10:23
time series data which is important for our work workflows so for example let's
10:26
our work workflows so for example let's
10:26
our work workflows so for example let's say if I want to see a forecast of the
10:29
say if I want to see a forecast of the
10:29
say if I want to see a forecast of the oil production so I'm basically
10:31
oil production so I'm basically
10:31
oil production so I'm basically interested in the trend set pattern
10:33
interested in the trend set pattern
10:33
interested in the trend set pattern because I want to apply the time series
10:35
because I want to apply the time series
10:35
because I want to apply the time series analysis which is basically a typical
10:37
analysis which is basically a typical
10:37
analysis which is basically a typical machine learning work workflow so it is
10:40
machine learning work workflow so it is
10:40
machine learning work workflow so it is only possible if your data is stored as
10:42
only possible if your data is stored as
10:42
only possible if your data is stored as a Time series manner so time series
10:45
a Time series manner so time series
10:45
a Time series manner so time series means it always has some temporality
10:47
means it always has some temporality
10:47
means it always has some temporality associated with it I mean it's
10:49
associated with it I mean it's
10:49
associated with it I mean it's chronological order maintain then this
10:52
chronological order maintain then this
10:52
chronological order maintain then this data could be a a periodic or sporic
10:56
data could be a a periodic or sporic
10:56
data could be a a periodic or sporic periodic means when we have some defined
10:59
periodic means when we have some defined
10:59
periodic means when we have some defined frequency like daily hourly or minute
11:01
frequency like daily hourly or minute
11:01
frequency like daily hourly or minute base or there are possible that there
11:04
base or there are possible that there
11:04
base or there are possible that there are some sporic operation for for
11:07
are some sporic operation for for
11:07
are some sporic operation for for example if there is some downtime
11:08
example if there is some downtime
11:08
example if there is some downtime activity in certain well so this
11:11
activity in certain well so this
11:11
activity in certain well so this downtime activity is is like a sporic
11:13
downtime activity is is like a sporic
11:13
downtime activity is is like a sporic event I mean there is no uh fixed
11:17
event I mean there is no uh fixed
11:17
event I mean there is no uh fixed interval associated with it the downtime
11:19
interval associated with it the downtime
11:19
interval associated with it the downtime started may may now it may finish maybe
11:22
started may may now it may finish maybe
11:22
started may may now it may finish maybe in hour or it may go for more than an
11:25
in hour or it may go for more than an
11:25
in hour or it may go for more than an hour also third type is why we we want
11:28
hour also third type is why we we want
11:28
hour also third type is why we we want to see the time series data because it
11:30
to see the time series data because it
11:30
to see the time series data because it has some pattern associated with it and
11:33
has some pattern associated with it and
11:33
has some pattern associated with it and we our workflows want to analyze these
11:35
we our workflows want to analyze these
11:35
we our workflows want to analyze these pattern and there is always a
11:37
pattern and there is always a
11:37
pattern and there is always a seasonality associated with time series
11:40
seasonality associated with time series
11:40
seasonality associated with time series data which involves pattern that repeat
11:43
data which involves pattern that repeat
11:43
data which involves pattern that repeat at the known interval it could be daily
11:44
at the known interval it could be daily
11:44
at the known interval it could be daily weekly or
11:46
weekly or
11:46
weekly or or so second type of data is basically
11:49
or so second type of data is basically
11:49
or so second type of data is basically structural data so to explain this data
11:53
structural data so to explain this data
11:53
structural data so to explain this data I just I would like to give you one
11:55
I just I would like to give you one
11:55
I just I would like to give you one example let's say if some oil and gas
11:59
example let's say if some oil and gas
11:59
example let's say if some oil and gas company has some oil well let's say in
12:03
company has some oil well let's say in
12:03
company has some oil well let's say in the Texas in Texas there is one city C
12:05
the Texas in Texas there is one city C
12:05
the Texas in Texas there is one city C Cam Cameron so you can think of it's
12:09
Cam Cameron so you can think of it's
12:09
Cam Cameron so you can think of it's like okay maybe in North America that
12:11
like okay maybe in North America that
12:11
like okay maybe in North America that oil and gas company may have different
12:13
oil and gas company may have different
12:13
oil and gas company may have different oil wells in different geographical lo
12:16
oil wells in different geographical lo
12:16
oil wells in different geographical lo location so they always start with the
12:19
location so they always start with the
12:19
location so they always start with the top route that is basically a a country
12:22
top route that is basically a a country
12:22
top route that is basically a a country then they will go to the state or they
12:25
then they will go to the state or they
12:25
then they will go to the state or they also they may create a a Zone area as
12:28
also they may create a a Zone area as
12:28
also they may create a a Zone area as well I mean they can combine multiple
12:31
well I mean they can combine multiple
12:31
well I mean they can combine multiple geography they can start calling it as a
12:33
geography they can start calling it as a
12:33
geography they can start calling it as a Zone and after that they they are
12:35
Zone and after that they they are
12:35
Zone and after that they they are basically reaching towards the geography
12:38
basically reaching towards the geography
12:38
basically reaching towards the geography where Oil Well is actually located so
12:40
where Oil Well is actually located so
12:40
where Oil Well is actually located so they always identify it using latitude
12:43
they always identify it using latitude
12:43
they always identify it using latitude and longitude of oil well so on the
12:47
and longitude of oil well so on the
12:47
and longitude of oil well so on the right hand side I'm basically showing
12:49
right hand side I'm basically showing
12:49
right hand side I'm basically showing this kind of data which is a structural
12:51
this kind of data which is a structural
12:51
this kind of data which is a structural data as you can see this is a a graph
12:55
data as you can see this is a a graph
12:55
data as you can see this is a a graph based structure so on the center you see
12:57
based structure so on the center you see
12:57
based structure so on the center you see a well is located and there are three
13:00
a well is located and there are three
13:00
a well is located and there are three kind of branches coming out from from
13:03
kind of branches coming out from from
13:03
kind of branches coming out from from that well one is in in orange color one
13:06
that well one is in in orange color one
13:06
that well one is in in orange color one is in red color yellow and green so
13:09
is in red color yellow and green so
13:09
is in red color yellow and green so these different branches are basically
13:10
these different branches are basically
13:10
these different branches are basically showing hierarchies where this well is
13:13
showing hierarchies where this well is
13:13
showing hierarchies where this well is attached so well is obviously attached
13:16
attached so well is obviously attached
13:16
attached so well is obviously attached to some field where we are basically
13:18
to some field where we are basically
13:18
to some field where we are basically doing the drilling operation that field
13:21
doing the drilling operation that field
13:21
doing the drilling operation that field is as exist in some state so that's why
13:25
is as exist in some state so that's why
13:25
is as exist in some state so that's why that one one line shows okay this oil
13:28
that one one line shows okay this oil
13:28
that one one line shows okay this oil well exist in this state then there is
13:31
well exist in this state then there is
13:31
well exist in this state then there is another hierarchy which is basically for
13:33
another hierarchy which is basically for
13:33
another hierarchy which is basically for patum
13:35
patum
13:35
patum engineers if they want to know the
13:37
engineers if they want to know the
13:37
engineers if they want to know the reservoir and Zone and completion so
13:39
reservoir and Zone and completion so
13:40
reservoir and Zone and completion so completion is another concept it is
13:43
completion is another concept it is
13:43
completion is another concept it is basically an area under the oil well
13:47
basically an area under the oil well
13:47
basically an area under the oil well from where Oil actually produced so that
13:49
from where Oil actually produced so that
13:49
from where Oil actually produced so that area is basically called com completion
13:53
area is basically called com completion
13:53
area is basically called com completion then I'm showing other hierarchy as well
13:55
then I'm showing other hierarchy as well
13:55
then I'm showing other hierarchy as well for example if there are tools installed
13:58
for example if there are tools installed
13:58
for example if there are tools installed so whenever oil is coming on the Earth
14:01
so whenever oil is coming on the Earth
14:01
so whenever oil is coming on the Earth surface then there are multiple tool
14:03
surface then there are multiple tool
14:03
surface then there are multiple tool installed we call it pump and flow flow
14:05
installed we call it pump and flow flow
14:05
installed we call it pump and flow flow lines and separators so that hierarchy
14:07
lines and separators so that hierarchy
14:07
lines and separators so that hierarchy basically shows okay this oil well tools
14:09
basically shows okay this oil well tools
14:09
basically shows okay this oil well tools are basically connected to the these
14:12
are basically connected to the these
14:12
are basically connected to the these subsurface components on the earth then
14:14
subsurface components on the earth then
14:14
subsurface components on the earth then there are other hierarchy like events as
14:16
there are other hierarchy like events as
14:16
there are other hierarchy like events as I said if it is a downtime event or some
14:20
I said if it is a downtime event or some
14:20
I said if it is a downtime event or some some something else they have their own
14:22
some something else they have their own
14:22
some something else they have their own hierarchy so and this structure is
14:25
hierarchy so and this structure is
14:25
hierarchy so and this structure is basically a a graph based structure so
14:27
basically a a graph based structure so
14:27
basically a a graph based structure so you can see what we want want to store
14:30
you can see what we want want to store
14:30
you can see what we want want to store okay the node ID itself uh and the edges
14:34
okay the node ID itself uh and the edges
14:34
okay the node ID itself uh and the edges and the started and ended so all these
14:37
and the started and ended so all these
14:37
and the started and ended so all these entities in this system are connected by
14:39
entities in this system are connected by
14:39
entities in this system are connected by several hierarchies and as I said okay
14:42
several hierarchies and as I said okay
14:42
several hierarchies and as I said okay the same entity could be attached to the
14:44
the same entity could be attached to the
14:44
the same entity could be attached to the multiple hierarchy as well so the main
14:47
multiple hierarchy as well so the main
14:47
multiple hierarchy as well so the main challeng is now we talked about the data
14:50
challeng is now we talked about the data
14:50
challeng is now we talked about the data the time series data and the structural
14:52
the time series data and the structural
14:52
the time series data and the structural data so time series data could be very
14:56
data so time series data could be very
14:56
data so time series data could be very huge it could be a PAB of data for more
15:01
huge it could be a PAB of data for more
15:01
huge it could be a PAB of data for more than 10 years or so and for a
15:04
than 10 years or so and for a
15:04
than 10 years or so and for a structural it is basically a graphical
15:08
structural it is basically a graphical
15:08
structural it is basically a graphical data so our challenge is to create a
15:11
data so our challenge is to create a
15:11
data so our challenge is to create a scalable cloud storage so that we can uh
15:15
scalable cloud storage so that we can uh
15:15
scalable cloud storage so that we can uh store this massive data and this graph
15:18
store this massive data and this graph
15:18
store this massive data and this graph structural data so how did we solve this
15:22
structural data so how did we solve this
15:22
structural data so how did we solve this so our first pro problem was the data
15:24
so our first pro problem was the data
15:24
so our first pro problem was the data injection problem so in the production
15:26
injection problem so in the production
15:26
injection problem so in the production operation as I said there are several s
15:29
operation as I said there are several s
15:29
operation as I said there are several s installed and there are several tools
15:31
installed and there are several tools
15:31
installed and there are several tools tools installed and we have different
15:33
tools installed and we have different
15:33
tools installed and we have different kind of data sources as well so there
15:35
kind of data sources as well so there
15:35
kind of data sources as well so there are typical data source like if there is
15:37
are typical data source like if there is
15:37
are typical data source like if there is a on
15:38
a on
15:38
a on Prem relational database that is called
15:41
Prem relational database that is called
15:41
Prem relational database that is called a production data management solution
15:43
a production data management solution
15:43
a production data management solution there could be Edge base or iot based
15:46
there could be Edge base or iot based
15:46
there could be Edge base or iot based devices streaming data there are mobile
15:49
devices streaming data there are mobile
15:49
devices streaming data there are mobile phone as well I mean someone is just
15:51
phone as well I mean someone is just
15:51
phone as well I mean someone is just capturing some data and preparing some
15:54
capturing some data and preparing some
15:54
capturing some data and preparing some report there could be a a physics based
15:57
report there could be a a physics based
15:57
report there could be a a physics based model if you guys don't about the physic
15:59
model if you guys don't about the physic
15:59
model if you guys don't about the physic physics based model so for
16:02
physics based model so for
16:02
physics based model so for example uh in any oil field they install
16:06
example uh in any oil field they install
16:06
example uh in any oil field they install multiphase flow meter so the job of this
16:09
multiphase flow meter so the job of this
16:09
multiphase flow meter so the job of this multiphase flow meter is like uh to
16:13
multiphase flow meter is like uh to
16:13
multiphase flow meter is like uh to check the uh
16:16
check the uh
16:16
check the uh liquid liquid characteristic and the
16:19
liquid liquid characteristic and the
16:19
liquid liquid characteristic and the heat exchange rate and uh What is the
16:22
heat exchange rate and uh What is the
16:22
heat exchange rate and uh What is the characteristic of the oil gas and water
16:26
characteristic of the oil gas and water
16:26
characteristic of the oil gas and water proportion when when whenever that Raw
16:28
proportion when when whenever that Raw
16:28
proportion when when whenever that Raw oil coming from the earth so we want to
16:32
oil coming from the earth so we want to
16:32
oil coming from the earth so we want to capture a different kind of data from
16:35
capture a different kind of data from
16:35
capture a different kind of data from the different kind of sources so we
16:37
the different kind of sources so we
16:37
the different kind of sources so we definitely need a scalable data
16:40
definitely need a scalable data
16:40
definitely need a scalable data injection system and data data is really
16:43
injection system and data data is really
16:43
injection system and data data is really important because uh 80% of the time any
16:46
important because uh 80% of the time any
16:46
important because uh 80% of the time any production engineer or petroleum
16:48
production engineer or petroleum
16:48
production engineer or petroleum engineer they always uh keep looking at
16:50
engineer they always uh keep looking at
16:50
engineer they always uh keep looking at data because their workflows are
16:53
data because their workflows are
16:53
data because their workflows are basically data
16:55
basically data
16:55
basically data intensives so to solve this problem what
16:58
intensives so to solve this problem what
16:58
intensives so to solve this problem what what we did we we created autonomous
17:01
what we did we we created autonomous
17:01
what we did we we created autonomous agent so autonomous agent is a very
17:03
agent so autonomous agent is a very
17:03
agent so autonomous agent is a very famous term these days we use it for for
17:07
famous term these days we use it for for
17:07
famous term these days we use it for for creating machine learning models and llm
17:09
creating machine learning models and llm
17:09
creating machine learning models and llm model also but yeah autonomous agent
17:12
model also but yeah autonomous agent
17:12
model also but yeah autonomous agent means anything any software process
17:15
means anything any software process
17:15
means anything any software process which is running continuously and acting
17:18
which is running continuously and acting
17:18
which is running continuously and acting on the events it receive so for us the
17:23
on the events it receive so for us the
17:23
on the events it receive so for us the we we always want okay some process to
17:25
we we always want okay some process to
17:25
we we always want okay some process to run closer to the data source and keep
17:28
run closer to the data source and keep
17:28
run closer to the data source and keep listening for the events like if some
17:30
listening for the events like if some
17:30
listening for the events like if some new data point arrive or if you want to
17:34
new data point arrive or if you want to
17:34
new data point arrive or if you want to fet the data for more than 20 years
17:37
fet the data for more than 20 years
17:37
fet the data for more than 20 years because there are few Wells who are
17:39
because there are few Wells who are
17:39
because there are few Wells who are producing oil since most since last more
17:42
producing oil since most since last more
17:42
producing oil since most since last more than 100 years as well so we want to
17:45
than 100 years as well so we want to
17:45
than 100 years as well so we want to fetch all this 100 Years of data and you
17:48
fetch all this 100 Years of data and you
17:49
fetch all this 100 Years of data and you can just imagine the scale if some well
17:52
can just imagine the scale if some well
17:52
can just imagine the scale if some well producing a second based frequency data
17:54
producing a second based frequency data
17:54
producing a second based frequency data for 100 years there could be billions of
17:57
for 100 years there could be billions of
17:57
for 100 years there could be billions of data points so we want to push all these
18:00
data points so we want to push all these
18:00
data points so we want to push all these data points to our cloud storage so we
18:03
data points to our cloud storage so we
18:03
data points to our cloud storage so we produce this agent this agents in simple
18:05
produce this agent this agents in simple
18:05
produce this agent this agents in simple term you can think of it's like a
18:07
term you can think of it's like a
18:07
term you can think of it's like a Windows service running continuously and
18:11
Windows service running continuously and
18:11
Windows service running continuously and uh whenever it receives some event from
18:13
uh whenever it receives some event from
18:13
uh whenever it receives some event from the on Prem data source it is uh pushing
18:16
the on Prem data source it is uh pushing
18:16
the on Prem data source it is uh pushing those data points securely to the cloud
18:18
those data points securely to the cloud
18:18
those data points securely to the cloud storage so each agent is
18:21
storage so each agent is
18:21
storage so each agent is basically uh associated with a unique
18:24
basically uh associated with a unique
18:24
basically uh associated with a unique agent ID what we handle on cloud and uh
18:28
agent ID what we handle on cloud and uh
18:28
agent ID what we handle on cloud and uh for secure
18:29
for secure
18:29
for secure if you want to secure the communication
18:31
if you want to secure the communication
18:31
if you want to secure the communication Channel between the onr and Cloud so we
18:34
Channel between the onr and Cloud so we
18:34
Channel between the onr and Cloud so we use a cloud cloudbased service account
18:37
use a cloud cloudbased service account
18:37
use a cloud cloudbased service account so that whenever someone is installing
18:39
so that whenever someone is installing
18:39
so that whenever someone is installing those agent that person is always using
18:42
those agent that person is always using
18:42
those agent that person is always using the encryption key coming from the
18:44
the encryption key coming from the
18:44
the encryption key coming from the cloudbased service account then that
18:46
cloudbased service account then that
18:46
cloudbased service account then that installer is basically decrypt that key
18:49
installer is basically decrypt that key
18:49
installer is basically decrypt that key and after that we don't have any idea
18:51
and after that we don't have any idea
18:51
and after that we don't have any idea about a de key because we are not
18:53
about a de key because we are not
18:54
about a de key because we are not storing anything and whenever I'm
18:56
storing anything and whenever I'm
18:56
storing anything and whenever I'm communicating from onr to Cloud using
18:59
communicating from onr to Cloud using
18:59
communicating from onr to Cloud using some messaging q that channel is also
19:02
some messaging q that channel is also
19:02
some messaging q that channel is also encrypted using same key so it makes
19:04
encrypted using same key so it makes
19:04
encrypted using same key so it makes sure that we are not spoofing anything
19:06
sure that we are not spoofing anything
19:06
sure that we are not spoofing anything and uh each agent is basically talking
19:09
and uh each agent is basically talking
19:09
and uh each agent is basically talking to the same messaging topic associated
19:12
to the same messaging topic associated
19:12
to the same messaging topic associated with its cloud service account so this
19:15
with its cloud service account so this
19:15
with its cloud service account so this is for handling the data injection so in
19:17
is for handling the data injection so in
19:17
is for handling the data injection so in this slide I'm basically showing the
19:19
this slide I'm basically showing the
19:19
this slide I'm basically showing the typical life cycle of of any agent and
19:22
typical life cycle of of any agent and
19:22
typical life cycle of of any agent and right hand side I'm also showing one
19:24
right hand side I'm also showing one
19:24
right hand side I'm also showing one picture so this picture is is basically
19:26
picture so this picture is is basically
19:26
picture so this picture is is basically from the patent I find Fed so this is
19:29
from the patent I find Fed so this is
19:29
from the patent I find Fed so this is this patent is already approved so this
19:32
this patent is already approved so this
19:32
this patent is already approved so this is for showing the life cycle of the
19:34
is for showing the life cycle of the
19:34
is for showing the life cycle of the agent so this basically shows any agent
19:37
agent so this basically shows any agent
19:37
agent so this basically shows any agent can send the time SE and structural data
19:39
can send the time SE and structural data
19:39
can send the time SE and structural data what I showed earlier we are using the
19:41
what I showed earlier we are using the
19:41
what I showed earlier we are using the Proto the Google Proto buff for format
19:44
Proto the Google Proto buff for format
19:45
Proto the Google Proto buff for format for serializing and desizing we can
19:47
for serializing and desizing we can
19:47
for serializing and desizing we can always send the incremental incremental
19:50
always send the incremental incremental
19:50
always send the incremental incremental data on daily basis or or overly basis
19:53
data on daily basis or or overly basis
19:53
data on daily basis or or overly basis or second base or historical and this
19:55
or second base or historical and this
19:55
or second base or historical and this agent can also request some
19:57
agent can also request some
19:57
agent can also request some configuration at the startup on
19:58
configuration at the startup on
19:58
configuration at the startup on periodically or demand from cloud and it
20:01
periodically or demand from cloud and it
20:01
periodically or demand from cloud and it always use some messaging uh or
20:04
always use some messaging uh or
20:04
always use some messaging uh or communication mechanism so that it can
20:05
communication mechanism so that it can
20:05
communication mechanism so that it can scale real well it can also receive some
20:09
scale real well it can also receive some
20:09
scale real well it can also receive some commands from from cloud okay let's say
20:12
commands from from cloud okay let's say
20:12
commands from from cloud okay let's say someone wants to fetch the last 8 years
20:14
someone wants to fetch the last 8 years
20:14
someone wants to fetch the last 8 years of data then that person can basically
20:17
of data then that person can basically
20:17
of data then that person can basically use some apis hosted on on cloud and it
20:20
use some apis hosted on on cloud and it
20:20
use some apis hosted on on cloud and it can start those jobs manually and uh
20:23
can start those jobs manually and uh
20:24
can start those jobs manually and uh other challenge what we solve okay we we
20:27
other challenge what we solve okay we we
20:27
other challenge what we solve okay we we need something to monitor so we are
20:29
need something to monitor so we are
20:29
need something to monitor so we are using the same uh uh messaging pop up
20:33
using the same uh uh messaging pop up
20:33
using the same uh uh messaging pop up communication mechanism to send the
20:35
communication mechanism to send the
20:35
communication mechanism to send the logging and monitoring status as well
20:37
logging and monitoring status as well
20:37
logging and monitoring status as well those we call hardbeat and we are
20:39
those we call hardbeat and we are
20:39
those we call hardbeat and we are monitoring those hardbeats on on cloud
20:42
monitoring those hardbeats on on cloud
20:42
monitoring those hardbeats on on cloud so we basically created some dashboard
20:45
so we basically created some dashboard
20:45
so we basically created some dashboard so that we always see okay our agent is
20:47
so that we always see okay our agent is
20:47
so that we always see okay our agent is up or
20:48
up or
20:48
up or not now moving over to the agent life
20:52
not now moving over to the agent life
20:52
not now moving over to the agent life cycle the typical agent life cycle start
20:54
cycle the typical agent life cycle start
20:54
cycle the typical agent life cycle start if someone registered that agent on
20:56
if someone registered that agent on
20:56
if someone registered that agent on cloud you using some set of apis so when
20:58
cloud you using some set of apis so when
20:59
cloud you using some set of apis so when agent is Agent is registered on cloud
21:02
agent is Agent is registered on cloud
21:02
agent is Agent is registered on cloud then we basically associate it with some
21:04
then we basically associate it with some
21:04
then we basically associate it with some cloud service account so we maintain the
21:06
cloud service account so we maintain the
21:06
cloud service account so we maintain the authentication and authorization part
21:08
authentication and authorization part
21:08
authentication and authorization part and that cloud service account is also
21:10
and that cloud service account is also
21:10
and that cloud service account is also associated with some Cloud pffs some
21:12
associated with some Cloud pffs some
21:12
associated with some Cloud pffs some mechanism so when I'm saying Cloud pups
21:14
mechanism so when I'm saying Cloud pups
21:14
mechanism so when I'm saying Cloud pups some mechanism so in G in Google Cloud
21:16
some mechanism so in G in Google Cloud
21:16
some mechanism so in G in Google Cloud it's a cloud psub or in Azure it is aure
21:20
it's a cloud psub or in Azure it is aure
21:20
it's a cloud psub or in Azure it is aure service bus and for AWS it is SNS or sqs
21:24
service bus and for AWS it is SNS or sqs
21:24
service bus and for AWS it is SNS or sqs and agent installation like after
21:26
and agent installation like after
21:26
and agent installation like after registering the agent I want to install
21:28
registering the agent I want to install
21:28
registering the agent I want to install it on the on Prem machine so that it is
21:30
it on the on Prem machine so that it is
21:30
it on the on Prem machine so that it is always attached to that particular data
21:32
always attached to that particular data
21:32
always attached to that particular data source so I use the same cloud service
21:35
source so I use the same cloud service
21:35
source so I use the same cloud service account key and after completing the
21:37
account key and after completing the
21:37
account key and after completing the installation if it is a Windows Server
21:39
installation if it is a Windows Server
21:39
installation if it is a Windows Server it basically encrypts it using the
21:41
it basically encrypts it using the
21:41
it basically encrypts it using the window data protection API so as soon as
21:44
window data protection API so as soon as
21:44
window data protection API so as soon as agent start it it basically started
21:46
agent start it it basically started
21:46
agent start it it basically started streaming data on cloud then the
21:48
streaming data on cloud then the
21:48
streaming data on cloud then the injection part is basically sending the
21:50
injection part is basically sending the
21:50
injection part is basically sending the oil production data from these oil field
21:52
oil production data from these oil field
21:52
oil production data from these oil field and data is pushed using the same
21:54
and data is pushed using the same
21:54
and data is pushed using the same messaging any pops of mechan
21:57
messaging any pops of mechan
21:57
messaging any pops of mechan mechanism and uh yeah we also can send
22:02
mechanism and uh yeah we also can send
22:02
mechanism and uh yeah we also can send commands from Cloud to onr for if we
22:04
commands from Cloud to onr for if we
22:04
commands from Cloud to onr for if we want to fetch any historical
22:09
want to fetch any historical
22:09
want to fetch any historical data so one important point point was
22:12
data so one important point point was
22:12
data so one important point point was let's say if there is a some disaster
22:15
let's say if there is a some disaster
22:15
let's say if there is a some disaster happens on cloud okay so what is the
22:17
happens on cloud okay so what is the
22:17
happens on cloud okay so what is the best way to reduce the downtime because
22:20
best way to reduce the downtime because
22:20
best way to reduce the downtime because I don't want our customer to wait so in
22:23
I don't want our customer to wait so in
22:23
I don't want our customer to wait so in disaster scenario the strategy we
22:25
disaster scenario the strategy we
22:25
disaster scenario the strategy we followed we created a global uh Cloud
22:29
followed we created a global uh Cloud
22:29
followed we created a global uh Cloud resource which is available in multis
22:31
resource which is available in multis
22:31
resource which is available in multis zone so whenever uh and also because we
22:34
zone so whenever uh and also because we
22:34
zone so whenever uh and also because we were using the cloud P pops up so by
22:37
were using the cloud P pops up so by
22:37
were using the cloud P pops up so by default Cloud pops up has the message
22:40
default Cloud pops up has the message
22:40
default Cloud pops up has the message storage for more than 7 days so is so
22:44
storage for more than 7 days so is so
22:44
storage for more than 7 days so is so even if some service is not available
22:47
even if some service is not available
22:47
even if some service is not available some disaster happen on cloud side so
22:50
some disaster happen on cloud side so
22:50
some disaster happen on cloud side so that message is still present in that
22:52
that message is still present in that
22:52
that message is still present in that que for at least 7 days and by creating
22:55
que for at least 7 days and by creating
22:55
que for at least 7 days and by creating a global resource that project is Global
22:59
a global resource that project is Global
22:59
a global resource that project is Global so it is available in the multiple Cloud
23:01
so it is available in the multiple Cloud
23:01
so it is available in the multiple Cloud zone so we can easily recover that
23:04
zone so we can easily recover that
23:04
zone so we can easily recover that disaster s scenarios and we can replay
23:08
disaster s scenarios and we can replay
23:08
disaster s scenarios and we can replay all those messages which are in the
23:10
all those messages which are in the
23:10
all those messages which are in the queue for last 7even days so it
23:12
queue for last 7even days so it
23:12
queue for last 7even days so it basically help us to achieve the
23:15
basically help us to achieve the
23:15
basically help us to achieve the disaster mechanism for solving the
23:17
disaster mechanism for solving the
23:17
disaster mechanism for solving the injection Pro problem now I want to
23:20
injection Pro problem now I want to
23:20
injection Pro problem now I want to store okay now data is arrived I I want
23:23
store okay now data is arrived I I want
23:23
store okay now data is arrived I I want to store that data on on cloud so the
23:26
to store that data on on cloud so the
23:26
to store that data on on cloud so the main challenge in any oil production
23:27
main challenge in any oil production
23:27
main challenge in any oil production operation is okay everyone start using
23:30
operation is okay everyone start using
23:30
operation is okay everyone start using their own terminology own data model so
23:33
their own terminology own data model so
23:33
their own terminology own data model so so someone has someone wants to call
23:36
so someone has someone wants to call
23:36
so someone has someone wants to call something like area Field Station or Val
23:39
something like area Field Station or Val
23:39
something like area Field Station or Val other person wants to call it some other
23:42
other person wants to call it some other
23:42
other person wants to call it some other tank and
23:43
tank and
23:43
tank and compressor so there are several domain
23:45
compressor so there are several domain
23:45
compressor so there are several domain concept like field asset surface
23:47
concept like field asset surface
23:47
concept like field asset surface subsurface equipment Val B holes and
23:50
subsurface equipment Val B holes and
23:50
subsurface equipment Val B holes and completion our goal is to present every
23:53
completion our goal is to present every
23:53
completion our goal is to present every single thing by using these three terms
23:55
single thing by using these three terms
23:56
single thing by using these three terms only one is called entities second is
23:58
only one is called entities second is
23:58
only one is called entities second is called properties and third one is
24:00
called properties and third one is
24:00
called properties and third one is called the relationships so entity is
24:01
called the relationships so entity is
24:01
called the relationships so entity is like anything it could be well or
24:03
like anything it could be well or
24:03
like anything it could be well or completer or any tool or compressor a
24:06
completer or any tool or compressor a
24:06
completer or any tool or compressor a properties could be any data coming from
24:10
properties could be any data coming from
24:10
properties could be any data coming from that particular entity it could be
24:11
that particular entity it could be
24:11
that particular entity it could be pressure temperature or oil flow rate
24:14
pressure temperature or oil flow rate
24:14
pressure temperature or oil flow rate and relationships is like how these
24:16
and relationships is like how these
24:16
and relationships is like how these hierarchies are attached to each other
24:18
hierarchies are attached to each other
24:18
hierarchies are attached to each other so we basically created a canonical
24:20
so we basically created a canonical
24:20
so we basically created a canonical domain model so that we can represent
24:23
domain model so that we can represent
24:23
domain model so that we can represent every single thing by using a common
24:25
every single thing by using a common
24:25
every single thing by using a common terms it will solve the pro problem what
24:28
terms it will solve the pro problem what
24:28
terms it will solve the pro problem what we earlier when every single application
24:30
we earlier when every single application
24:30
we earlier when every single application was using their own data model and we
24:33
was using their own data model and we
24:33
was using their own data model and we are always solving this problem again
24:35
are always solving this problem again
24:35
are always solving this problem again and
24:37
and
24:37
and again so another use case just to
24:40
again so another use case just to
24:40
again so another use case just to explain you why we need a canical data
24:43
explain you why we need a canical data
24:43
explain you why we need a canical data model let's say in there some field
24:45
model let's say in there some field
24:45
model let's say in there some field there are two sensor installed one is
24:47
there are two sensor installed one is
24:47
there are two sensor installed one is streaming at both are streaming
24:49
streaming at both are streaming
24:49
streaming at both are streaming frequency one is coming from the iot
24:51
frequency one is coming from the iot
24:51
frequency one is coming from the iot source the second one is coming from
24:53
source the second one is coming from
24:53
source the second one is coming from some database let's say one one person
24:55
some database let's say one one person
24:55
some database let's say one one person calling is iot Source like esp.
24:58
calling is iot Source like esp.
24:58
calling is iot Source like esp. frequency in Herz and second person is
25:01
frequency in Herz and second person is
25:01
frequency in Herz and second person is calling just like ESP
25:03
calling just like ESP
25:03
calling just like ESP underscore uncore frequency in a in Herz
25:07
underscore uncore frequency in a in Herz
25:07
underscore uncore frequency in a in Herz but for our production domain model it
25:09
but for our production domain model it
25:09
but for our production domain model it is just ESP frequency we don't care
25:12
is just ESP frequency we don't care
25:12
is just ESP frequency we don't care about how how the main data source want
25:15
about how how the main data source want
25:15
about how how the main data source want to name it or tag it whenever data is
25:18
to name it or tag it whenever data is
25:18
to name it or tag it whenever data is coming to our data storage we want to
25:20
coming to our data storage we want to
25:20
coming to our data storage we want to apply our own domain model concept so it
25:23
apply our own domain model concept so it
25:23
apply our own domain model concept so it basically help us to resolve the
25:26
basically help us to resolve the
25:26
basically help us to resolve the different terminology and data standard
25:28
different terminology and data standard
25:28
different terminology and data standard because now we understand okay we are
25:30
because now we understand okay we are
25:30
because now we understand okay we are talking a same language and we are not
25:32
talking a same language and we are not
25:32
talking a same language and we are not confused about what other people are
25:35
confused about what other people are
25:35
confused about what other people are giving some other some other tag or
25:39
giving some other some other tag or
25:39
giving some other some other tag or name so other challenge was okay we were
25:43
name so other challenge was okay we were
25:43
name so other challenge was okay we were we wanted to store this
25:46
we wanted to store this
25:46
we wanted to store this huge amounts of Time s data instruction
25:49
huge amounts of Time s data instruction
25:49
huge amounts of Time s data instruction data and our main requirement was we
25:52
data and our main requirement was we
25:52
data and our main requirement was we always want to keep the history of data
25:54
always want to keep the history of data
25:54
always want to keep the history of data because let's say if I'm running a
25:56
because let's say if I'm running a
25:56
because let's say if I'm running a forecasting operation or I'm running
25:58
forecasting operation or I'm running
25:58
forecasting operation or I'm running some uh other recommendation in your
26:01
some uh other recommendation in your
26:01
some uh other recommendation in your workflow I wanted to see the history of
26:03
workflow I wanted to see the history of
26:03
workflow I wanted to see the history of data always I don't want to delete
26:06
data always I don't want to delete
26:06
data always I don't want to delete anything I always want to upend
26:08
anything I always want to upend
26:08
anything I always want to upend everything so we solve this problem
26:10
everything so we solve this problem
26:10
everything so we solve this problem using Boral storage so Boral so B
26:13
using Boral storage so Boral so B
26:13
using Boral storage so Boral so B temporality is a concept when we assign
26:16
temporality is a concept when we assign
26:16
temporality is a concept when we assign at least two times 10 one is like a
26:18
at least two times 10 one is like a
26:18
at least two times 10 one is like a valid time second one is called the
26:20
valid time second one is called the
26:20
valid time second one is called the transaction time so valid time is like
26:22
transaction time so valid time is like
26:22
transaction time so valid time is like the actual time when that physical
26:24
the actual time when that physical
26:24
the actual time when that physical measurement happen at the source like
26:27
measurement happen at the source like
26:27
measurement happen at the source like when we make the pressure at source and
26:30
when we make the pressure at source and
26:30
when we make the pressure at source and proection time is when I basically
26:33
proection time is when I basically
26:33
proection time is when I basically storing that data into our Cloud stories
26:37
storing that data into our Cloud stories
26:37
storing that data into our Cloud stories so it basically helped us by storing
26:39
so it basically helped us by storing
26:39
so it basically helped us by storing these two time stamp it help us to uh
26:42
these two time stamp it help us to uh
26:43
these two time stamp it help us to uh run the temporal queries it is very
26:45
run the temporal queries it is very
26:45
run the temporal queries it is very useful for running any kind of
26:47
useful for running any kind of
26:47
useful for running any kind of historical analysis it also help help us
26:51
historical analysis it also help help us
26:51
historical analysis it also help help us to achieve the temporal Trends and
26:53
to achieve the temporal Trends and
26:54
to achieve the temporal Trends and running the long running immutable Cal
26:57
running the long running immutable Cal
26:57
running the long running immutable Cal calculation
26:59
calculation
26:59
calculation and uh we can Al we can always run the
27:03
and uh we can Al we can always run the
27:03
and uh we can Al we can always run the different data points between different
27:05
different data points between different
27:05
different data points between different data versions so that we
27:08
data versions so that we
27:08
data versions so that we can see the complete history of the data
27:11
can see the complete history of the data
27:11
can see the complete history of the data and it was really important for our
27:13
and it was really important for our
27:13
and it was really important for our historical work workflows so on right
27:15
historical work workflows so on right
27:15
historical work workflows so on right hand side as you can see I'm just
27:17
hand side as you can see I'm just
27:17
hand side as you can see I'm just showing you you the examples on the top
27:20
showing you you the examples on the top
27:20
showing you you the examples on the top it is showing the time series storage
27:21
it is showing the time series storage
27:21
it is showing the time series storage where I'm assigning two time stamp one
27:25
where I'm assigning two time stamp one
27:25
where I'm assigning two time stamp one is on the row level other one is at
27:27
is on the row level other one is at
27:27
is on the row level other one is at column LEL
27:28
column LEL
27:28
column LEL and Below we I'm showing the structural
27:31
and Below we I'm showing the structural
27:31
and Below we I'm showing the structural data where the two times St is basically
27:34
data where the two times St is basically
27:34
data where the two times St is basically installed so on the first time St I'm
27:37
installed so on the first time St I'm
27:37
installed so on the first time St I'm showing that okay well2 was not attached
27:39
showing that okay well2 was not attached
27:39
showing that okay well2 was not attached to battery one but after some time on
27:42
to battery one but after some time on
27:42
to battery one but after some time on 13th October after one one day the well
27:45
13th October after one one day the well
27:45
13th October after one one day the well is attached to the battery well so I
27:47
is attached to the battery well so I
27:47
is attached to the battery well so I always wanted to store the history of
27:50
always wanted to store the history of
27:50
always wanted to store the history of data and how we achieved this we
27:52
data and how we achieved this we
27:52
data and how we achieved this we basically built a by temporary storage
27:55
basically built a by temporary storage
27:55
basically built a by temporary storage and just to give you one example okay
27:58
and just to give you one example okay
27:58
and just to give you one example okay uh so we are basically uh we started
28:02
uh so we are basically uh we started
28:02
uh so we are basically uh we started with a cloud native version of it but
28:05
with a cloud native version of it but
28:05
with a cloud native version of it but slowly we move towards a cloud agnostic
28:08
slowly we move towards a cloud agnostic
28:08
slowly we move towards a cloud agnostic one so our first version was created
28:11
one so our first version was created
28:11
one so our first version was created using big table as a storage so on the
28:13
using big table as a storage so on the
28:13
using big table as a storage so on the top you are basically seeing the actual
28:15
top you are basically seeing the actual
28:15
top you are basically seeing the actual big table schema and bottom for the
28:17
big table schema and bottom for the
28:17
big table schema and bottom for the structural storage we use another
28:20
structural storage we use another
28:20
structural storage we use another database that's called datomic so
28:22
database that's called datomic so
28:22
database that's called datomic so datomic is basically by default provides
28:26
datomic is basically by default provides
28:26
datomic is basically by default provides a feature to run a temporal queries
28:28
a feature to run a temporal queries
28:29
a feature to run a temporal queries because it basically stores the version
28:30
because it basically stores the version
28:30
because it basically stores the version time and the trans and the transaction
28:35
time and the trans and the transaction
28:35
time and the trans and the transaction time so consumption worklow okay now our
28:38
time so consumption worklow okay now our
28:38
time so consumption worklow okay now our data is stored we have everything now we
28:40
data is stored we have everything now we
28:40
data is stored we have everything now we want to run a robust model so that we
28:43
want to run a robust model so that we
28:43
want to run a robust model so that we can uh uh create our consumption so our
28:45
can uh uh create our consumption so our
28:45
can uh uh create our consumption so our consumption workflows are basically
28:47
consumption workflows are basically
28:47
consumption workflows are basically interested in looking for the entities
28:49
interested in looking for the entities
28:49
interested in looking for the entities associate properties what I explained
28:51
associate properties what I explained
28:51
associate properties what I explained earlier using the same canical model I
28:54
earlier using the same canical model I
28:54
earlier using the same canical model I want to Traverse that Boral graph
28:58
want to Traverse that Boral graph
28:58
want to Traverse that Boral graph so that I can know okay if even I need
29:01
so that I can know okay if even I need
29:01
so that I can know okay if even I need to go multiple lbel down and find okay
29:05
to go multiple lbel down and find okay
29:05
to go multiple lbel down and find okay my pressure is coming from some flow
29:07
my pressure is coming from some flow
29:07
my pressure is coming from some flow Point location under certain Val so I
29:09
Point location under certain Val so I
29:09
Point location under certain Val so I can do that traversing as well I'm also
29:12
can do that traversing as well I'm also
29:12
can do that traversing as well I'm also interested in doing the lot of
29:15
interested in doing the lot of
29:15
interested in doing the lot of calculation on the time space data which
29:17
calculation on the time space data which
29:17
calculation on the time space data which is basically aggregation consuming and
29:19
is basically aggregation consuming and
29:19
is basically aggregation consuming and write back I want to apply the data
29:21
write back I want to apply the data
29:21
write back I want to apply the data quality attribute and all this data is
29:24
quality attribute and all this data is
29:24
quality attribute and all this data is basically feed into some calculation
29:26
basically feed into some calculation
29:26
basically feed into some calculation engine so that this calculation engine
29:28
engine so that this calculation engine
29:28
engine so that this calculation engine is basically running in the background
29:30
is basically running in the background
29:31
is basically running in the background and giving some recommendation and
29:32
and giving some recommendation and
29:32
and giving some recommendation and insight to production engineer okay and
29:34
insight to production engineer okay and
29:34
insight to production engineer okay and helping them to understand why their oil
29:37
helping them to understand why their oil
29:37
helping them to understand why their oil production is low or why their
29:38
production is low or why their
29:38
production is low or why their forecasting is is not is not matching so
29:42
forecasting is is not is not matching so
29:42
forecasting is is not is not matching so all this data is basically helping to
29:44
all this data is basically helping to
29:44
all this data is basically helping to run this consumption and work workflow I
29:46
run this consumption and work workflow I
29:46
run this consumption and work workflow I would like to give uh uh some example
29:50
would like to give uh uh some example
29:50
would like to give uh uh some example like I'm showing here some Advanced
29:52
like I'm showing here some Advanced
29:52
like I'm showing here some Advanced calculation like validity and S
29:55
calculation like validity and S
29:55
calculation like validity and S selection so validity means now we
29:58
selection so validity means now we
29:58
selection so validity means now we stored all data all time series data I
30:01
stored all data all time series data I
30:01
stored all data all time series data I want to see uh if there is a gap in my
30:05
want to see uh if there is a gap in my
30:05
want to see uh if there is a gap in my data so how much how much tolerance of
30:09
data so how much how much tolerance of
30:09
data so how much how much tolerance of this data Gap I can I can have in my
30:12
this data Gap I can I can have in my
30:12
this data Gap I can I can have in my work workflows so this calculation is
30:15
work workflows so this calculation is
30:15
work workflows so this calculation is basically let's say when I'm feeding
30:17
basically let's say when I'm feeding
30:17
basically let's say when I'm feeding this data to that calculation engine and
30:19
this data to that calculation engine and
30:19
this data to that calculation engine and that calculation engine sees some Gap so
30:22
that calculation engine sees some Gap so
30:22
that calculation engine sees some Gap so it basically a back fill to some status
30:26
it basically a back fill to some status
30:26
it basically a back fill to some status code so that engineer understand okay I
30:29
code so that engineer understand okay I
30:29
code so that engineer understand okay I have some bad status code or some
30:31
have some bad status code or some
30:31
have some bad status code or some unavailable status code so I mean it
30:34
unavailable status code so I mean it
30:34
unavailable status code so I mean it sound simple but it's a difficult
30:36
sound simple but it's a difficult
30:36
sound simple but it's a difficult problem to solve because just imagine I
30:39
problem to solve because just imagine I
30:39
problem to solve because just imagine I someone is seeing last 8 years of second
30:41
someone is seeing last 8 years of second
30:41
someone is seeing last 8 years of second based frequency data so I always need to
30:44
based frequency data so I always need to
30:44
based frequency data so I always need to find the last known good data point so
30:48
find the last known good data point so
30:48
find the last known good data point so that I know okay until that point I need
30:51
that I know okay until that point I need
30:51
that I know okay until that point I need to back fill so this data point could be
30:54
to back fill so this data point could be
30:54
to back fill so this data point could be maybe 1 month ago or could be 25 years
30:57
maybe 1 month ago or could be 25 years
30:57
maybe 1 month ago or could be 25 years ago as well also I want to know okay I
31:00
ago as well also I want to know okay I
31:00
ago as well also I want to know okay I want to merge the different data streams
31:03
want to merge the different data streams
31:03
want to merge the different data streams from different data source for
31:06
from different data source for
31:06
from different data source for example any single oil field if there is
31:10
example any single oil field if there is
31:10
example any single oil field if there is some uh uh Edge device installed and
31:14
some uh uh Edge device installed and
31:14
some uh uh Edge device installed and someone is also streaming data from
31:16
someone is also streaming data from
31:16
someone is also streaming data from Mobile by capturing some image or
31:19
Mobile by capturing some image or
31:19
Mobile by capturing some image or anything so but it is possible that both
31:22
anything so but it is possible that both
31:22
anything so but it is possible that both are sending the same kind of a stream so
31:25
are sending the same kind of a stream so
31:25
are sending the same kind of a stream so I need some way to know okay these are
31:28
I need some way to know okay these are
31:28
I need some way to know okay these are same kind of data so I need to D
31:30
same kind of data so I need to D
31:30
same kind of data so I need to D duplicate it and I want to merge it to
31:32
duplicate it and I want to merge it to
31:32
duplicate it and I want to merge it to the sing sing Single stream
31:35
the sing sing Single stream
31:35
the sing sing Single stream other calculations are like okay uh
31:40
other calculations are like okay uh
31:40
other calculations are like okay uh there could be a monteo simulation there
31:43
there could be a monteo simulation there
31:43
there could be a monteo simulation there could be a joining of Time series it
31:46
could be a joining of Time series it
31:46
could be a joining of Time series it could be more than two time time series
31:48
could be more than two time time series
31:48
could be more than two time time series so overall we created around, 1500 time
31:52
so overall we created around, 1500 time
31:52
so overall we created around, 1500 time series calculation what we are feeding
31:54
series calculation what we are feeding
31:55
series calculation what we are feeding it to generate these insights
31:59
and yeah I would like to give some
32:00
and yeah I would like to give some
32:01
and yeah I would like to give some interesting numbers so this application
32:03
interesting numbers so this application
32:03
interesting numbers so this application is basically uh deployed around the
32:06
is basically uh deployed around the
32:06
is basically uh deployed around the globe mostly for the oil and gas company
32:08
globe mostly for the oil and gas company
32:08
globe mostly for the oil and gas company in South America in Southeast Asia and
32:10
in South America in Southeast Asia and
32:10
in South America in Southeast Asia and Middle East it is managing more than
32:13
Middle East it is managing more than
32:13
Middle East it is managing more than 20,000 oil wells we have 1500 1500 plus
32:18
20,000 oil wells we have 1500 1500 plus
32:18
20,000 oil wells we have 1500 1500 plus time calculation so I would like to give
32:21
time calculation so I would like to give
32:21
time calculation so I would like to give you some numbers for from one client
32:24
you some numbers for from one client
32:24
you some numbers for from one client that client has 7,500 oil well the
32:27
that client has 7,500 oil well the
32:27
that client has 7,500 oil well the typical hierarchy was started from
32:29
typical hierarchy was started from
32:29
typical hierarchy was started from company some labels then field then then
32:32
company some labels then field then then
32:32
company some labels then field then then well and number of entities around
32:34
well and number of entities around
32:34
well and number of entities around 30,000 total relationship be served
32:37
30,000 total relationship be served
32:37
30,000 total relationship be served 60,000 and properties are 1.5 billion
32:41
60,000 and properties are 1.5 billion
32:41
60,000 and properties are 1.5 billion this client has the data streaming for
32:44
this client has the data streaming for
32:44
this client has the data streaming for 25 years so the typical calculations so
32:47
25 years so the typical calculations so
32:47
25 years so the typical calculations so shows we injested on 14 billion data
32:51
shows we injested on 14 billion data
32:51
shows we injested on 14 billion data points and yeah these are other numbers
32:53
points and yeah these are other numbers
32:53
points and yeah these are other numbers like uh the achievements I mean we able
32:56
like uh the achievements I mean we able
32:56
like uh the achievements I mean we able to achieve the 98% time saving 88% cost
32:59
to achieve the 98% time saving 88% cost
32:59
to achieve the 98% time saving 88% cost saving 80% reduction of the data
33:02
saving 80% reduction of the data
33:02
saving 80% reduction of the data preparation to time we able to achieve
33:05
preparation to time we able to achieve
33:05
preparation to time we able to achieve the well up time as well and the typical
33:07
the well up time as well and the typical
33:07
the well up time as well and the typical tag stag I would like to talk okay this
33:10
tag stag I would like to talk okay this
33:10
tag stag I would like to talk okay this is purely a microservice based
33:12
is purely a microservice based
33:12
is purely a microservice based architecture we have 40 plus
33:14
architecture we have 40 plus
33:14
architecture we have 40 plus microservices we are running over two
33:17
microservices we are running over two
33:17
microservices we are running over two commun cluster and uh 90% services are
33:22
commun cluster and uh 90% services are
33:22
commun cluster and uh 90% services are written in Escala few services are
33:24
written in Escala few services are
33:24
written in Escala few services are written in go and Python and we are
33:27
written in go and Python and we are
33:27
written in go and Python and we are using AKA a lot AKA is a actor based
33:30
using AKA a lot AKA is a actor based
33:30
using AKA a lot AKA is a actor based framework and uh yeah everything is
33:33
framework and uh yeah everything is
33:33
framework and uh yeah everything is because we started with the cloud native
33:36
because we started with the cloud native
33:36
because we started with the cloud native but still our future goal was Cloud
33:38
but still our future goal was Cloud
33:38
but still our future goal was Cloud agnostic so we by default started using
33:40
agnostic so we by default started using
33:40
agnostic so we by default started using cuties so we are using cuties since 2015
33:44
cuties so we are using cuties since 2015
33:44
cuties so we are using cuties since 2015 I mean it's very early days and for
33:46
I mean it's very early days and for
33:46
I mean it's very early days and for storage we are using red is post datomic
33:49
storage we are using red is post datomic
33:49
storage we are using red is post datomic and big table so yeah I think so that's
33:52
and big table so yeah I think so that's
33:52
and big table so yeah I think so that's all I have and I'm open for any question
34:00
[Music]