Sramana Mitra: Your use case is quite broad-based. You’re focused on the telecom vertical, but the use cases are broad. It’s not productized. You’re providing professional services style of use cases on top of a common data handling platform.
Prashant Kumar: Yes. We are like SAP for telcos. The platform is agnostic to any vertical. The approach we have taken is very SAP-oriented. I’ll give you an example. If United Airlines buys Amtrak SAP’s procurement system, the underneath tech doesn’t change.
Sramana Mitra: We know this well. We deal with product positioning all the time. Let’s go into the platform. You have chosen to position it within telecom across various use cases. When you get into a telecom customer, there are various use cases that you’re going after with your platform but building the functional logic on top of it through your professional services. That’s the architecture of the company, right?
Prashant Kumar: Right.
Sramana Mitra: What’s in the platform?
Prashant Kumar: Just a note on the telecom side, that’s just something we started growing. Most of us started in telecom. We are slowly rolling into non-telcos.
Sramana Mitra: It’s a common go-to-market strategy in platform companies where you go into the market with one vertical and you start broadening into other verticals.
Prashant Kumar: The platform is cloud native. The reason for that is because of the scalability issues. Our original architecture comes from Netflix’s open source architecture of what they were doing on AWS. We took that and we figured out how we can adopt it for the analytics workload. We’ve got two different components within our platform.
Sramana Mitra: Are you agnostic to databases?
Prashant Kumar: Our DataOps, which was built to ingest billions of events at stream, is agnostic and super scalable. Our biggest deployment is running on 100 nodes, which has been running every day for the last five years.
Sramana Mitra: On what database?
Prashant Kumar: It’s all in-memory. We use our own tech and system components that we have used to combine it together to scale out the data.
Sramana Mitra: You said something about a particular client that is very large and you said something about which database you were using.
Prashant Kumar: I was saying that our own platform is running on a hundred nodes. The end point there is Hadoop. Everything that we do from injection to everything is in-memory.
Sramana Mitra: You have an in-memory architecture for your processing that draws from a database agnostic source. The customer has you draw from there, process it in-memory, and then do what the customer is asking for on top of that.
Prashant Kumar: Database can be a source, but when you look at IoT, it comes directly from the equipment. For example in an airport, multiple sensors would be deployed. They can stream directly. Database can also be a source and it does sometimes, but most of our data comes from streams.