MariaDB Foundation Sea Lion Champions Nominees: Sumit Srivastava

Interview with Sumit Srivastava, nominated in the Adoption & Industry Impact category.

I had the pleasure of speaking with Sumit Srivastava, SVP Business Development & Products at Tayana, a Bangalore-based telecom software company. Sumit was nominated for championing MariaDB adoption in the telecommunications industry and its demonstration of open-source database delivering performance, scalability, and business impact in mission-critical systems. Like me, Sumit sits between technical and business — “the business people think I’m technical, and the technical people think I’m business,” as he put it. It made a lively conversation.

The Interview

Kaj: Sumit, Congratulations on your nomination. Tell me about Tayana and your role there.

Sumit: I started as a database administrator in 2002 — Oracle and MySQL, thereafter more than 15 databases over the span of 25 years. At Tayana, I run business development and product management. We are a 350 people strong product company with predominant focus on Telecom products Design and development. Tayana is a wholly owned subsidiary of Jupiter capital group. Our company is focused on trust, excellence, quality, long-term partnerships with customers, and value creation.

Kaj: You serve telecom operators across the EMEA, Asia Pacific and Americas. What makes telecom particularly demanding for a database?

Sumit: Our solutions are business-critical and require high performance, resilience, scalability, and reliability. You get less than 15 milliseconds to respond. Think about it: when you connect your phone to a network, it must latch in milliseconds, otherwise radio resources are wasted. When you make a chargeable call, the system debits your account every 60 seconds. If I don’t get the response back in time, your call drops. When you pay with a card at an airport, the session must be created and confirmed before the queue behind you grows. The moment I’m not getting the response back, TPS degrades, and within 15 to 20 seconds, the whole system stops.

Kaj: Before MariaDB, how did you handle this?

Sumit: Three separate databases. One in-memory store for the session layer, where data needs sub-millisecond access. A relational database for the persistent transactional data. And a separate analytics platform for reporting. Three skill sets, three deployments, three support contracts, three license renewals. Too much complexity.

Kaj: Why MariaDB specifically?

Sumit: I spent significant time evaluating our database choices against ten critical parameters. The most important insight was this: many people in the industry think MariaDB is just a MySQL replacement. That thinking is wrong. MariaDB started there in 2008 — I know that. But right now it has created its own niche. It is a paradigm shift and it fits all of our purposes.

What I saw was a database marketplace. One platform — in-memory, RDBMS, high-performance columnar analytics. GridGain for ultra-low latency, ColumnStore for massive parallel processing, Galera for clustering. What I liked about the acquisitions is the growth mindset: not acquire and kill, like some large database companies do. MariaDB acquires and collaborates, embeds the best technology, and creates an ecosystem.

Stability matters to me too. I have seen analytics platforms change owners three times in a few years — and when your infrastructure keeps changing hands, you cannot make the long-term commitments your customers depend on. MariaDB started in 2008 with the courage to fight for independence. That mindset has held.

Kaj: What have you achieved so far, and what is next?

Sumit: Our online charging system runs today on MariaDB, serving customers up to 2 million subscribers. When we started, it was running at 200 TPS. After some tuning, 300. After the MariaDB India team visited us on-site, the same platform reached 850 TPS. But that is not the stopping point.

My objective for the next 100 days: scale to 50 million subscribers. The architecture is three layers — GridGain as the in-memory session store. As it supports ANSI SQL, so we don’t need to rewrite the application queries; MariaDB InnoDB for persistent transactional data; ColumnStore for analytics and CDR queries. With that stack, I expect 40,000 to 50,000 transactions per second.

Kaj: And AI?

Sumit: We are realistic. We build things that actually solve problems. Two are live already. First: a multilingual AI chatbot for call centers in the EMEA region, where travelers speak Russian, French, Arabic, Bengali — the operator cannot train staff for every language. Our AI detects the preferred language and handles the conversation, translating in real time if escalated to a human agent. It is not replacing anyone. It is helping agents understand customers who would otherwise feel lost.

Second: AI analytics on every call, not a sample. Transcript, sentiment, tone, issue classification — we call it coaching. The output feeds back to the app, so recurring billing problems eventually become self-service resolutions before the customer calls at all.

But the database is what makes it possible. When the AI agent fires a live query, I need the response before the user notices a pause. Without a database fast enough and integrated enough, the AI experience collapses.

Kaj: Any message for the MariaDB community?

Sumit: Change the market perception. MariaDB is not for small applications, not only for open-source enthusiasts, not only a MySQL replacement. It is a marketplace with all the flavours from small to large enterprises. In the last five to eight years, MariaDB have crossed the parent [1]. The child has surpassed. Now let’s prove it at tier-one scale — together.

[1] Interviewer’s note: I had to look up “cross the parent”. It turns out to be an Indian idiom meaning that a child has grown beyond its parent — a mark of the highest respect. Sumit is saying that MariaDB, born as a MySQL fork in 2008, has long since become greater than its origin.