Bala Kutchibhotla - Destined to Redefine Databases


JASON: (intro)

BALA: 
Databases are a system where you can store data and retrieve.

JASON:
Databases have existed ever since writing was invented. With early computers it was stored on punchcards and magnetic tape. The modern computer database emerged in the 1960s and business adoption soon followed.

BALA:
Traditionally, databases are related to more of a transactional properties. That means that when you're going to the bank and you're debiting and crediting and all that stuff, there needs to be a record of what you heard done. There is no ambiguity down the road that you did it and then bank needs to, or the money to you, they should give it back. So there has to be a system where they need to record these transactions and that's the way database is a primary use case occured. Accounting was the first use case.

JASON:
Kutchibhotla says there are four properties of a database. He uses the acronym acid, shich stands for atomicity, consistency, isolation, and durability.

BALA:
That means that you cannot afford to lose the data. Let's say you put $10,000 in the bank, that transaction is committed forever. If I take $10,000 but not credit the other account, then there is a problem. Those are what we call the relational databases with certain type of what I call acid properties: atomicity consistency, isolation, durability. That means that you cannot say that, “Hey, I know what the transaction is committed, but I don't know. I can't commit it. I don't know.” You cannot say that for that database.

JASON:
He was born in India and grew up amid the development of the country’s IT boom. He says being an engineer — a good engineer — is something many aspire to in a very competitive landscape. 

BALA:
I love computer science for sure and more so databases. And that's how I've been traveling for the last 20 years with databases. All the way from my master thesis, which is an object database kernel. And that's how I got into my first job in India. By the way, the company was managed by Ben Trainer who runs the Google cloud and then from there moved to Oracle about 14 years back and then doing all the database management stuff there.

JASON:
Databases are in Kutchibhotla’s DNA. He thinks a lot about the future of the technology and lately his focus has been on virtualizing databases.

BALA:
Virtualization was a trend about fourteen years back. But everyone moved to cloud and database virtualization has not fully taken off. So you need cloud like simplicity.

JASON:
Virtualizing a database requires more than slapping one into the compute node.

BALA:
Why do you do virtualization, right? Like to consolidate. But the struggle that I see is databases need to be virtualized at the data level to really reap the benefits of virtualization. If you create more database compute nodes, but the data is not virtualized, your pain points are moving towards data. So when you're trying to deal with cloning of the data, but just because you have compute node virtualization but the databases are not virtualization then you're actually creating bigger problems. When we're talking about honest database virtualization, the innovation has to happen on both sides. The compute side as well as the data side.

JASON:
This question of innovating on the data side is one reason Kutchibhotla decided to bring his vision to Nutanix. The company made its name simplifying storage and making infrastructure invisible. He wants to revolutionize database ops in a similar way and create a product release using the cloud. The project is called ERA.

BALA:
ERA I would say a true database virtualization to start with and then extending it to more like cloud like experience for database as service. So we kind bring the true database. virtualization when I said true database, virtualization and how do you do snapshots, how do you manage your data, not just running your database servers in your virtualized compute environment.

JASON:
There are four core challenges that database virtualization needs to address: provisioning proliferation data protection and patching. What follows is Kutchibhotla’s explanation in a keynote at the Nutanix NEXT conference in Copenhagen last year.

BALA:
Provisioning is one of the biggest pain point that it takes like three to five weeks to kind of even create a database for some of our big customers and it kind of spans across multiple teams. The second problem that I see is the proliferation, right? Proliferation is like creating copies of your databases. This is like our copy paste problem like I don't see anyone, I don't think anyone in this audience has never done a copy paste on their laptop. Right? Same thing happens for databases in your data center. Only thing is this seemingly simple innocuos operation of our daily life becomes the most dreaded, complex, long running error prone operation. I've seen customers spending like three days and having a dedicated job profile for cloning databases and then lineage is even more bigger pain for that. The third point is data production. If you're talking about business critical apps and databases, mission critical databases, it's an absolute must that you must help your ability to ability to go back to any point in time to protect your data and it could be malicious data. Someone, someone corrupted the data. It could be your own Butterfinger that drove the data kind of stuff. Your ability to protect data is absolute must and the last one is the patching pain that we talk about it. If you are managing databases like more than a hundred databases kind of stuff, and even a few dozens. It's an absolute must. You how to have complaints and standardization to how the database are under control kind of stuff.

JASON:
Kutchibhotla talks about provisioning, which used to take days that now can be done in minutes, as well as cloning and refreshing at will.

BALA:
Our innovation through era, not just a compute level virtualization of the databases, but the data level virtualization, the ability to clone at zero byte warhead and keeping the same quality of service for those clones. And we also took the problems that the database team must have gone through and tried to see how the virtualization platform can be morphed to solve those big problems. So you will see that one click patching. Just imagine a workload patching was a big pain for a lot of financial customers. For us thousands of databases. We could do it with one click today. So that's there. I see the honest database virtualization with the cloud experience getting blended and together, that's what we call ERA.

JASON:
Perhaps database virtualization could be put under the umbrella of the concept of simplicity. One feature Kutchibhotla envisions: being able to do database operations on an iPhone.

BALA:
People are moving to more the consumer experience, whether it is the iPhone or smartphones. Why should the DB ops people not enjoy the consumer grade experience? At the end of the day, you don't need to worry about databases. They should be simple. We don't need to worry about managing them. They should be taken for granted, let's put it that way. That may be the better word. I would expect database as a service, like how we enter the power in this room, we're seeing the lights, right? It's a service. We do not know who generates it, who process it, how it gets distributed to us, but we just enjoy it. That's how we use it. Database should be like that too. They are not the end. They're a means to an end.

JASON:
Bala Kuchibhotla is the vice president, general manager and founder of Nutanix Era. 
Previously he was VP and architect at Oracle overseeing Oracle’s private cloud initiative.
This is the Tech Barometer Podcast, I’m Jason Lopez.
Tech Barometer is produced by the online news magazine, The Forecast.
For more stories and podcasts you can find them at the-forecast-by-Nutanix dot com.