Databases for AI: How Agents Reshape Database Operations

Speaker A: Foreign. If you think about AI, AI is nothing but running on top of data. So data I would say is critical to running any successful AI initiative or application. And on the other hand you also need AI to manage the database. Most of your data or critical data for a company is stored in uh, your database, a variety of databases, right? And then how do you really operate those databases seamlessly, uh, and at scale? You need AI to do that. So it's kind of interrelated. You know, you could always run open source small databases on containers, uh, in a stateless manner. But now you can run stateful databases on containers. That means you can run more complex and mission critical applications on it. So that's the first thing that's happened. The second thing is, as everybody knows, is the whole excitement around AI and the innovations that's taking place in AI and that how AI is helping run and manage these databases in a more efficient manner. So that really is a very exciting area where people, we can build agents to manage your databases, to run your databases, to optimize your databases and to get school squeeze out more performance, more scale and reduce cost. So customers are choosing to run databases on multiple public clouds along with on premises data centers. And they want simplicity in terms of how they manage this entire complex environment. So I would say those are the big three trends that have really emerged and are starting to accelerate in the market today. And we really transformed our entire development life cycle process, uh, to adapt to using AI so that we can actually build product faster and deliver more functionality to our customers. So that is happening there. The second aspect is how do you actually use AI within the platform that is exposed to our customers? So how can our customers now take advantage of AI and really benefit from productivity gains, cost savings, uh, or whichever uh, benefit they're looking for. In that case we start, we are looking at exposing new capabilities within the platform that makes our customers lives easier. So where you needed a lot of manual intervention, uh, we're trying to see if we can automate a lot of that or give you an assistant. So this would be more like, I would say a driving assistant to say, look, I need to understand why my databases are uh, not performing up to par, what could be slowing it down? What are the threat vectors that I should be looking out for, uh, uh, how do I automate the lifecycle management of certain databases, uh, how do I create a quick clone uh, of my database, uh, based on certain triggers that may be happening in my business. So all of those things we can create agents for that and then give our customers the ability to deploy those agents and to really benefit from the agentic architecture. Today, uh, as we are kind of experimenting with this technology and we're developing that in our own labs, what we're seeing is you can actually run agents to do any kind of uh, tasks, uh, for these databases. You can uh, provision your databases, you can backup, recover your databases, you can actually uh, apply patches to it. Uh, you can do cloning, you, you can set those up. So agents pretty much can do a, uh, majority of the operations uh, of a database today. In addition to that you get some really interesting insights. You can monitor these databases more efficiently. You can know which database is not performing up to your target spec or what is really slowing it down. Is it an indexing issue, is it a query optimization issue? You can actually make the databases more secure. Uh, what are the vulnerabilities that are going to be exposed around this database? So a lot of this interesting stuff now the agents can really enable you to do that. And as we are building these technology, we've seen that within our own uh, lab environments that you can become a lot more efficient with agents. What would take you, I would say months can now be done in a matter of days or weeks overall. So that is the kind of efficiency gains that we are seeing by moving to an agentic architecture. An application can be broken down into components and each one of these components can be built independently. That's what typically is called a microservice. Now each one of these microservices can uh, run on its own set of databases, for example, and what you really want, and those microservices, typically you want to bring them up and down really quickly. This is where containers really come in very handy and you want to reuse them as well. So the idea is if you actually have a microservices that typically is deployed on the container environment, right? So if that's the case, you want your databases to be very efficiently run on containers. And that's what we're making it possible. The other thing is you want the ability to spin them up and down really quickly. And um, in a VM environment it takes a little bit longer to do that. In a container you can do that very efficiently. Also container nicely packages that entire microservice, uh, in a self contained environment in some ways that makes it very easy to become portable. So you can port that. You want to run this on premises today, you want to run it on the cloud tomorrow, or run another different server on premises. It's very easy to just take that container and deploy it somewhere else versus you having to move the entire VM so it becomes bulky, containers are lightweight. So all these things really make containers more efficient uh, and makes it really possible to build and assemble applications quickly, run these applications very quickly and to manage them more efficiently. So that's the reason why customers are choosing to deploy applications on a containerized environment, uh, for their different kinds of workloads. If you're thinking about migrating, let's say from a uh, legacy VM environment to a more modern environment, typically what's the number one workload that's running on that legacy VM environment? It tends to be a database. It may be an application on top of it, but databases are your number one or the obviously the heavy duty workloads and the most mission critical workloads. So when you're thinking about moving from a legacy environment to a modern environment, there are multiple ways you can do that. The one way uh, that customers gravitate towards typically is lift and shift. So I take the VM and I just lift it and shift it to the modern vm. Right. Uh, modern platform. Now what are the advantages? You can do that in certain cases, but in other cases where you have databases running and these databases are configured in, that's a high availability cluster or a disaster recovery cluster or failover. Uh, uh, in a setup what happens is when you do a lift and shift you lose context of these VMs, the database VMs. So if you're in a cluster, it doesn't recognize that it was in a cluster. So it'll take the first vm, move it to uh, the modern environment, it'll take the second VM of the cluster and move it to the modern environment, but doesn't know how the first one and the current VM are interconnected. This is called application consistent movements or backups, uh, and restores. You can't do that. Uh, so customers spend a considerable amount of time once they migrated individual VMs to now correlate them to recreate that previous environment of a high availability cluster or a disaster recovery cluster or failover. That takes a lot of time and effort to do that because now you're stitching together all these VMs without really having context. Now we have a better solution. We say with Nutanix database service you uh, can actually connect to your databases on the legacy VM in whatever configuration they're set up. Single instance, high availability disaster recovery failover. It backs up those environments, maintaining context. So it does application consistent snapshots then you can create a replica or you can create a standby on the modern uh, environment like Nutanix and then easily be able to replicate the entire uh, database as it was without losing a beat. So without having significant downtime, you moved your most sophisticated database configurations and environments onto nutanix without disruption to your business. That is huge. And you can be confident that when you have cut over to nutanix, those applications will work as they were intended, which is a big deal. This typically takes a very long time. With ndb we are able to reduce that time by up to 82%. So it's a significant acceleration in terms of migrations, giving you the confidence as well that the new environment will work for you as intended and the majority of these cases as well. We see a performance improvement of 2x on these databases. So without paying anything extra, doing anything extra, what you're getting is 2x performance improvement, 82% faster acceleration, up to 82% faster acceleration and all the benefits of a modern platform that nutanix delivers to you. So it's a huge, huge value add for our customers. And you can do this whether you're moving it from an on premise to an on premise environment or from on premises to a cloud, or if you're running it in the cloud to another cloud or back to on premises. So we support all these environments with the same APIs and the same process that you would use in any kind of system. Overall, if you look at a NEO cloud, why would I use a NEO cloud? I'm using it for running my AI workloads, right? So I need access to GPUs, I need an AI stack, uh, to build my next generation AI application. One of the things as I'm building my AI workload is I need access to vector databases and uh, data that's stored in a vector table. So what does NEO Cloud need to offer? They need to offer choice to their customers to use uh, from a variety of vector databases, uh, so the customer can pick and choose what they want. If they want Postgres for vector capabilities or Mongo or a pure play like Pinecone or Milvus, uh, NEO cloud should be able to offer that in the backend. Now each one of these databases require uh, lifecycle management. They need to be provisioned, they need to be scaled, backed up. It gives them recovery capabilities, patched, cloned, upgrade. So either NEO clouds have to build this on their own or they need to automate this using a platform like ndb, where regardless of what database they run, they get all the capabilities and operations taken care of, uh, by a single platform. And that's what NDB offers. So that's why there's a powerful combination of NEO Cloud plus ndb. And I would say this is the reason Chronoscale has partnered with Nutanix. Is our, uh, entire stack that we offer for them to give them the ability to give them a modern platform for their end users, along with all the data capabilities.