The cloud allows complete freedom of processes to the enterprise. It provides complete scalability to small enterprise and startups while ensures efficiency and agility to established companies. As one of the biggest providers, AWS is a behemoth with the heart of a startup- fast, excited and always focusing on partners and customers’’ interests. Pranam Chatterjee, CEO, BluePi talks about the today and tomorrow of AWS support and his own ambitions for BluePi.
- How does cloud channelize digital transformation for enterprises? How do you enable your customers to access advanced technologies like Machine learning, AI, analytics and virtual reality through the cloud?
The biggest advantage with the cloud is ease of access. It’s all pretty much available. It’s almost like the democratization of IT. Previously you needed a lot of skill to be able to provision hardware, you’d be needing to do sizing. The cost of failure was significantly high as well, because once you procure there is a long period of gestation that you go through before you acquire the hardware, and once you have done it, you’ve done it.
Here the biggest advantage is, you start small. A massive amount of usage is coming from startup segment.
Whether it is AIML, serverless computing or containerization, Amazon is the largest behemoth, they are the largest but act and behave like a startup. So a lot of innovation is now essentially being made readily available to other IT organizations of varying sizes which they can readily go and adopt. So, for example, when AWS came up with Lambda which is like serverless computing, it probably took the competitors 3-6 months to follow suit.
The whole point is, it’s a beautiful concept. It completely takes away your maintenance overhead. There were two distinct divisions in enterprise historically- there were IT Infra Management division and the DevelopmentEcosystem. So the developers would build the solution, and then the deployment and management would be handed over the wall to the IT folks. Because the barrier to infrastructure has come down so low, the same teams can also do it now. And the containerization that has happened, what you build in development is going to happen in production. There is going to be no difference whatsoever.
The environment issues have completely gone away. So I think, the increasing influence of this new tech is decreasing barriers to adoption for the industry of any size.
This transformation also is happening, a lot of these large enterprises have a very good understanding so they have built homegrown or third-party solutions around their business problems, but they are stillmuch focused on their delivery capabilities. They need someone to help them in their cloud adoption journey and on the way, a lot of them are trying to modernize. So it’s like killing two birds with the same stone.So you are trying to modernize your infrastructure and at the same time, modernize the applicationsalso to take advantage of these new innovations that cloud is providing. The pace of innovation is just massive, right now.
So I think languages, JAVA came out somewhere around 1999 and became popular, it’s been around about 20 years now. In the last 4-5 years if you look at just the programing language systems, SWIFT has come out, you have SCALA, and Python, the pace of innovation is just massive.
A plethora of innovation is just happening and people are choosing from such a massively diverse system for solving their problems and are able to stitch them together because the innovations have brought down barriers and that has become that much easier.
- How do you position AWS, Microsoft and Google cloud solutions for your customers? Please provide your views on the benefits of the AWS platform since you have chosen it?
The truth of the matter is, when we started 4-5 years back, AWS was the only platform available. The second advantage is, because they have been there for such a long time, the maturity of the offerings is significant. So you will never walk into a customer conversation,and he will never ask “is AWS reliable? Is it stable? That conversation never happens because the whole host of success stories out there to talk about.
So that’s the first big story out there.
Second is, its very developer friendly. I will give you a classic example. We are doing the implementation of a large-scale data warehouse, which is typically a load shift- you load it overnight and you do visualization on putting on top of it next day. There’s a delay. So we are trying to build something which will be real time and on a very very massive scale for one of the biggest logistics providers in India. Now, while doing so, we encountered the challenge that the scaling of the platform wasn’t happening as we expected it.
The response time you see from Amazon, the customer obsession that they have and partner observation is just mind-blowing. Our problem becomes their problem, so overnight they will come, help you and ensure you find a solution so that’s very helpful aswell.
Another important part is, having worked with all three providers, the AWS partner program is the most focused on partner success.
This I think is the biggest differentiators.
- Tell us something about the architecture of a Data Lake on AWS and the particular advantages it offers for storage, processing and security?
Let’s take Data Lake piece from an architecture standpoint. It’s a very simple solution. It has Amazon S3, which is file storage, and you put data on top of it irrespective of whether it is structured or not. What is important is that point you need to figure out if you can out some semantic structure to it, so it is a set of data points you need to correlate and meet across different data sources. But since that is all the technology there is, there is no expensive Hadoop cluster you need to run, so absolutely zero investment from customer’s standpoint – there is no investment for technology. At the end of the day it is just filed storage. And then the sophistication comes on top of it. You have S3 and you put compute on top of it and there are different ways of doing that. If I am looking for ad-hoc queries I can put an Amazon Athena. If I am looking for something more sophisticated for a24*7 kind of use case, I will use anAmazon Redshift Spectrum. Or if I am more interested in something like a large-scale solution, I would use another solution.
The solution here is absolutely brilliant, because from a customer standpoint it addresses all the pain points that you have. So the breadth of the solution is out there.Because the data is posted in a structured format, you now have the ability to run AI-ML on top of it. So this single, homogenous data store ability that’s out there is obviously what people are referring to as Data Lake in the market today. But the single biggest advantage that comes through is of S3 and its simplicity. S3 was the first service that AWS offered. The maturity of that service grew by leaps and bounds for every customer.
And then, very veryfine-grained security constructs- because most of the infamous security breach stories you hear on AWS has been on the S3. And that has probably prompted AWS to provide very fine-grained controls on the data structures so that you can absolutely bulletproof your security constructson top of it.
- What are the challenges you face while migrating infrastructure to cloud? What are the change management issues customers face while migrating to the cloud? What in your opinion is the best way to handle it?
There are two kinds of customers that we see. The first is the ‘born in the cloud startup’unicorn and is probably co-lo, but now wants to move to the cloud. So there is a lot of talent and a lot of understanding of infrastructure as well.
Then there is enterprises who are very traditional and probably have reached 70- 80% level of virtualization, and now want to move to the cloud.
I think the problem statements are very specific to these two types of scenarios.In the first scenario, it’s more about education about what values AWS services can offer,
A lot of customers, because of their comfort in DB operations, would want to set up on a relational database serve. But we don’t want to do that because if we move them on to a relational database service, 80% of the management overhead just goes away. The backups are taken care of, security is configurable, so there it’s more about education.
In an org which is a little traditional, the investment that it takes from a partner about what the cloud can offer, is probably much more significant than what they need to do for a startup. It’s a disruption, the senior management makes a decision, and then it has a ripple effect across the organization. So somebody who is providing monitoring support also gets impacted.
So how do you bring this entire chain of command and how you enable them to adopt and make their life easy? The moment you start showing them how each specific feature will make their lives easier, it becomes a very quick win. But there is a significant amount of spending before that happens, to bring them up to speed about their abilities.
- Server- less environments are the future of enterprise architecture. How do you think the Indian enterprises are equipped to handle this?
There is a lot of desire. We are doing an implementation for a very large QSR where we are building their entire authentication and authorization platform on AWS which is a completely driven by serverless technology. I think there is appreciation and understanding of the value that it brings to the table.
It would still take some time for them to get up to speed. There are still challenges- like the containers, how do you work around that. So if you have a lot of peaks and lows, the containers will actually go down and it takes a bit of time for them to come up.
Second is, there are multiple languages you can program. You can choose what language you work in, becauseSDKs are available in pretty much all the languages. Depending on the use case, some specific language may be better suited. So those are some of the pieces that you need to be able to take care of. This actually ties in very well with the microservicesarchitecture as well. Soit’s about technology adoption, so if an organization is at a point where they have an understanding as to why they want to decouple a behemoth, a ball of wax kind of architecture, then that adoption becomes much faster.
In India we typically jump generations. It is like there is no landline phone, but most people have mobile phones. So that’s what we will start seeing very soon. A lot of organization while makinga move, will start embracing serverless rather than traditional architecture.
In fact we have customers who are very focused, when they are moving their workload, they are actually saying they want to do containerization, they want to go to Amazon Elastic Container Service (ECS), they want to go to an Uber network kind of construction rather than go to E2 instances. So I think we will very soon see a deluge there.
- What are the specific IT challenges faced by media and entertainment customers? How do you address those challenges with your bouquet of solutions and services? Please talk about your piStats solution for media and entertainment customers?
Media is a very different kind of thing, because the load is very unpredictable. A big story breaks online, then probably there is 50 times the load coming in. In some cases it is predictable, like when we have the election the result coming out, you can prepare for the deluge of data that will follow as the first story breaks. But when something like the cross-border strike that India did happens, that was absolutely out of the blue. But the traffic that generated was just enormous.
I think this kind of a workload is suited for cloud, primarily because you can deploy a very very small base technology layer and then, when scaling needs to happen, you can pretty much do it based on metrics. Very interestingly, when we moved ABP to AWS, the primary reasons were reliability and agility.
So wherever it would take them days to make the changes in the site design and framework, that time has shrunk significantly. We have been working with them for 3 years, every year the cost of operations is going down while the viewership is increasing. So you are delivering more with less, because all the news services are coming in and transforming their business.So ABP is all serverless now.
A bit about piStats now. It has four different components. One is the CMS, which essentially allows a very intelligent workflow. You plan a story, then there is collaboration built into it, so you can take notes, move the story to a copy edit stage, and then through the whole lifecycle. In fact it’s almost like a task board. Then the second piece is the automation of the home and section pages- there are typically three types of pages you have on a publishing site- a Home page, a section page and an article page. In the Home or section page you have a 3, 4 or 5 column layout and you put the story up in that.
The stories that go in there are now driven by logic. Earlier, an editor would sit and say which stories would go in the top, but right now, we can say the most viewed story in the last 30 minutes, needs to be here. So there is intelligent logic built into the entire platform which automatically makes the whole thing dynamic. How do we do it? There is an analytics piece on top of it. That takes all the information from every user who pulls every bit of information about what they do on the site and the mobile app. We do personalization on top of it, we see what exactly each individual has done on the site, and also get historical data and see what he is looking at, and we will tailor the content only as per his interest. So there is analytics then personalization, and that’s what the product is about.
We now have APIs, so we can now pull them and show them, so the editor now has a view right there in the CMS about the kind of stories that are trending.
One of the features that we are going to launch over the next one month is what we call competitive analysis. It picks up all the social media stories that have been pushed by the competitors, brings them into a single screen, does some NLP on it and breaks it down by topic to figure out which publication actually took the story out first, whose story is doing the best on social media. Then you have a sense of who broke the story and what did I do about it etc.
As regards the product, we are struggling a bit with giving it the due focus it requires, it needs to be forked out as a separate entity on its own. Today, there are three customers who are using it.
The one thing that I didn’t touch upon – piStat is available in probably 6 Indian languages till now. So once you publish a story, you figure out the key words for it and the right tags for it to trend on a social media platform- automated tagging etc, which works across 6 Indian languages.
So the whole idea is that it is a solution that can scale and work for a lot of publishers.
- What are your top focus areas over the next 2-3 years? What role does your current vendor partnerships play in your future plans?
We are totally committedto AWS, in the sense that we are an advanced partner right now and will come to premier partnership pretty soon. We have done Big Data competency, DevOps competency and migration competency already. From a business standpoint, there are certain things we stay away from, we don’t want to do run of the mill work that a lot of other shops are doing. We want to bring value to our customers in technology areas which really unlock some business value.
AIML is a big focus for us. We are building a system for a fairly large retail player that helps them optimize their retail chain. The idea is that the new services that AWS rolls out – how we take those services to the market as quickly as possible so the businesses we are working with can benefit from this new tech and innovate really really quickly.
- What are your takeawaysfrom this Summit?
I had some very interesting comments- right now the market is exploding. In the next 2-3 years, the opportunity will boom, it’s a 10x growth that we will see. People who have adopted technology have already done it. Now they will look at moving up the value chain. So we will see a lot of adoption coming in, that’s one thing that’s very obvious.
The second thing is that the continued commitment we see and the audacity with which you walk into a customer meeting and look at a customer statement or challenge, you can look at them and say, Yes, we can deliver.
So as a group, the idea is you keep raising the bar and setting customer expectation in a manner which is not conceivable.From a group-wise perspective, it’s important to take piStats to as many customers as possible. The other focus is to bring AI ML to as many people as possible in the Indian market.