STL TechTalk | Edge Computing | All Things Tech
I'm Manish Sinha, the CMO at STL, Sterlite Technologies
and today we talk about something that all of us know a little
bit about and most of us want to know a lot more about.
What is it? Edge Compute.
Well...I'm sure you've heard the term and if you're watching this video
on LinkedIn or twitter or any time when you watch videos on
Netflix or Amazon Prime or whatever is your favorite streaming
service, you know, you're getting it from something from
a server that's not too far away.
They are sending the stuff close to you.
They are sending the things that you need for your internet experience
close to the edge.
That's the Basic understanding, but how about we try to dive
into it a little bit more. Today we have with us Sandeep Dhingra
who has been a technologist with IBM and Cisco and now he
is the Chief Technology officer at the Netflix services and
software business at STL.
Welcome Sandeep. Thank You Manish! Glad to be here and I believe
you have described Edge pretty well.
So one of the things that I actually want to start with Sandeep,
you know, why should we care about Edge Computing? Very good question,
right and especially during the time when there's a lot of
hype around it.
So I will say three things, number one
It is a major generational architecture shift, right. The 2nd it
is essential for the key revolution that is going on
in the digital infrastructure called Internet of Things.
It is not IoT, but there are huge number of use cases which
will be enabled by Egde Compute.
So, for these three things it is very important.
So, let me talk about it.
What do I mean by architectural shift?
If you look at in 1960s, we had this big mainframes, right?
We had all the terminals we connect to i.e. the
standard centralized model, right? And then in 80s kind of everything
was upside down and we moved to a client-server model
and the AS400 version and you know Oracle servers.
These were the type of things that were Kings... and
then in 2000 and again, we took a turn and we started Cloud Computing.
Now Cloud Computing is you know it was a large-scale
model, shared infrastructure across this thing, but it is centralized.
Now you have seen how it has revolutionized
IT and technology in itself.
But now, we realize when the IoT came, we realize that you
know, everything cannot be and must not be run in the cloud,
i.e. centralized model.
So, then what did we do in Edge Computing, cloud moves to the
edge, the computing moves to the Edge, but now it completes
Now if you look at IDC, they have displayed it very nicely.
It is really a generational shift that is completing the
full technology cycle from centralized to distributed combined.
So that is about the architecture shift.
Now, there are 10, 20, 30, 40, 50 billions of devices are going to
be connected in future.
But the point is a huge amount of data will not only be created
at the Edge, but it will be processed, needed to be computed and processed
at the edge and by some estimates
we are talking about 45 to 50 percent of overall data that
needs to be processed at the edge. So that is what is changing.
That is why we should care.
That is very very interesting and thanks for taking us through
the journey, but two things that really popped out to me
is that this whole thing of Edge Computing has been here
for nearly two decades and also that way soon it will take
over 40 to 50 percent of the things that we are Computing
But one thing that I would love to know from you is what
exactly is Edge Computing?
So, there is lot of confusion about Edge, right?
So, first thing that we realize that the users, the devices
like cell phones, your laptop, your sensors, your robotic arms,
potentially these things connect.
That is the physical location is called Edge.
Now if you talk about Edge Computing, Edge Computing is
nothing but computing happening on the edge, that basically
a topology element. So, compute can happen at the cloud at the
This gives me about you know, what is happening at the edge. Now
cloud computing is really a style of computing that
basically, provides you, shared infrastructure, the resource elasticity, you can
up and down, you can provide through a platform provides
like horizontal scale,
it provides you catalog pay-as-you go model.
So, these are the basic style of computing. When you create the style
of computing to edge of the network.
You can call it Edge cloud or cloud model.
Thanks for telling us a little bit about what exactly it is
and I think the differentiation between Edge and Computing
on the edge and the cloud, all of that really helps. If you can
help us understand, why is Edge Compute so disruptive?
What is it fundamentally changing?
Let's say like, what does Edge Computing do? Right.
Let's say if you do this amount of compute of work at the edge
of the network where things are connecting.
So, it does provide three things. Number one.
It provides a very low latency that enables lots of applications.
That would not have been possible otherwise.
Then it ensures when you have huge amount of data
that is being generated that does not need to be sent back
to the cloud or data center.
Third. There is, it helps keep the data local. If there was no compute
at the edge and computation was necessary to make decisions,
Then, the data would have to be inevitably sent to a data
center or cloud
but with edge compute, with local processing, we can keep the data local.
These three things are enabled by The Edge that
basically makes it possible for thousands of
net new applications to be run and new use cases form that
a new efficiencies and productivity in, you know hundreds of different
verticals in sectors.
That is why it is disruptive.
Yeah, it is very interesting. Thousands of new applications
running but you actually mentioned something that probably
is a very simple word in the way technologists understand,
but I wanted to understand it in a more simpler way, which
is latency. What is really latency mean and you know, what are
some of the reasons and why do we care too much about it?
Really, literally it means time it takes for you know,
information or packet from point A to point B, right?
So, typically it is an Intense of basic and so, but what is
happening with IoT is there is a lot of machine-to-machine
communication. So, at machine scale, the machines when we
we talking microseconds,
Let's say if there is a gas pump in a refinery that is monitoring
a pressure that needs to be shut quickly without human and human
intervention and you know, lots of these things can be adjusted
now that requires ultra-low latency. Very very low latency.
The latency in human scale and human applications is different
than in machine scale, human scale it is tens of milliseconds
in machine scale,
it is under 10, under 1 and sometimes in microseconds.
Anything else that you think is making it imperative for the Edge
compute to become important?
So yes, there are other things, one is the sheer amount of
data, right. When you have millions of sensors or you have
you know, 8K streaming surveillance videos, that data you don’t
want to send it because it is costly to send data.
So, just the bandwidth and data is another one, right. And then there are scenarios
where you know, you need autonomy.
That is, you need Cloud, but you need to have most of their
first just to be able to run if there is no Cloud. You basically
use cloud only to send let's say, data for further analysis
or you send for training your machine learning models.
In the case of let's say a hospital or a military outpost,
where you still make everything, needs to work properly and
then you have the privacy and security
for data localization.
So, these are the key imperatives that you know, make Edge Computing
really important and critical.
On the other hand, Sandeep.
We also know that Edge Computing ranges from a lot of different
use cases. Right.
Maybe you talked about machine-to-machine, there is a wide range
I also hear that,
it's extremely important for Edge Computing to be at its best
for things like connected cars and autonomous vehicles.
What is happening there?
Why is that important?
So, there are two majors. One with the edge does not have any
impact like whatever we are doing now, If I am doing a web
application. It's running without the edge, I am fine.
But some applications get
enhanced with the edge that you
touch in the beginning, right? But then applications that are really... make use of Edge
in a super enhanced way that is where we talk about like say
high definition video surveillance from
4K 8K video streams.
But the real difference that comes where there are applications
that cannot run without the edge, that are called edge native applications.
There is no way they can run without the edge,
and autonomous cars
you mentioned, vehicles you mentioned, is one such example.
One use case that is also very important is Telco space,
its called multi-access Edge compute.
Thank you so much for sharing all of this with us Sandeep
and looking forward to meeting you at the edge.
Thank you Manish,
I have to say we are on the Edge of a revolution.
Thank you, Manish thanks for having me.
Thanks so much.