Transcript
Hi, Mike Matchett with Small World Big Data. I'm here today talking about another advance in cloud storage. Conceptually, we're bringing block services to the cloud. We can not just get object storage and file storage, but now we can get high performance block storage with some pretty cool solutions. I've got LightBits Labs here today. We're going to talk with Kamsky. Just hold on a second. So welcome, Kam. Welcome to our show. Thank you Mike. Great to be here. So you know this is pretty interesting solution you've got here bringing in from a kind of a unique perspective. It's not just another file system, another cloud object storage system, but we're actually bringing some high performance, what we might consider sand kind of storage into the cloud. How did you just first tell us how did you get pulled into this? How did you get involved in the storage industry, and how did you get into light beds? You know, way back a long time ago, we at a previous company built the very first NVMe controller in the industry when NVMe as a protocol first came about. And that kind of pulled me into storage. I've been in compute storage, networking for many years. And not just storage, but high performance storage and eventually with a focus on the cloud. So kind of bringing together some of the shifts that were happening in the industry back then, it was a move to the cloud and building the right protocols to support high performance storage as well as, you know, taking advantage of the full performance of flash. And it all came together when we built when we started Light Bits in 2016 to build the most cost efficient, high performance cloud storage. And that is entirely software defined and designed for the cloud. So that's how it all started. And, you know, working with Victor Willans and Aaron Kerzner and the other founders, we started the company in 2016. All right. So you've been around, you know, seven years. It looks like if I'm if I'm doing the math right, time to get a few versions of this out and make sure that it's a scalable and got a few enterprise customers under your belt. So we've heard that we're very happy with this. You've now extended this from being something that might initially have been a software defined solution you'd bring into the enterprise, into the data center, to something that you'd run in colo's private colos to now being run in public clouds. How does how does that work? Does it does it work the same in all those different environments? Yeah, that's a great question. You know, we we designed our solution, the core product for the cloud and for high performance. And initially the customers that we signed up were all on prem private clouds. These are enterprises in financial services and e-commerce that are building their own clouds for internal use, building services for other teams within the company to use. And and they're looking for software defined everything, a very agile infrastructure that supports multiple applications. That's where we started. Then we added to that tier two cloud service providers such as cloud. Cloud is using us. Fits is using US, which is a financial cloud service provider, as well as others in the US that are building specialized cloud services. And they need high performance block storage. More recently, we took the same solution and ported it to AWS and Azure. And I'm very excited about our solution with Azure, which just came out in September. And it allows a customer to get the same high performance block storage solution in their data centers on prem, as well as on Azure or AWS. Same experience, same cost efficiency. And it allows them also to buy capacity from little bits and choose to deploy that either in their private data centers or on the public cloud, and have that flexibility to migrate between the two as they can. Shift their capacity that they've licensed as they shift with their workloads. So let me just let me just ask you this. When think of cloud service providers, they have native block storage. And and when you're when you're going into these cloud providers, what are some of the key things people should be thinking about what you offer as an advantage versus the native block storage that's already there? Yeah, that's a great question. There are lots of storage options available on the public cloud. So, you know, what problem are we trying to solve? If you look at the native cloud storage solutions that are out there, they address a good chunk of the market, you know, sort of the low hanging fruit of moving. Lots of workloads that have already moved to the public cloud are able to address their storage needs with what's available today on the public cloud. However, there are these mission critical applications that tend to be more storage intensive, performance intensive, and especially latency sensitive that are having a hard time moving to the public cloud. Why? Because they have storage performance bottlenecks that cannot be addressed with the existing solutions on the public cloud, or if they can be addressed, they become very expensive because they have to go with the highest performance storage tier, which charges not just for capacity, but also for IOPs and maybe for some data services. So the cost just becomes prohibitive. And that's preventing many of these workloads. A lot of them are database applications from moving to the public cloud like Oracle, PostgreSQL, MongoDB, Kubernetes, environment, VMware environment where it's still latency sensitive. That's where we come in. So our solution is designed for storage intensive workloads, where you want to be able to scale your storage independently from compute and achieve performance. That's similar to local flash. So one of the first things we did as a company early on was several years ago, was we defined a new protocol, NVMe over TCP, took it to the market, got it standardized with with other partners such as Intel, Dell, Cisco and others who are all investors in markets. And and it's part of NVMe fabrics. And then we built the entire storage solution around it. That includes resiliency. We can do three zone synchronous replication either on prem or on the public cloud. So this is a true enterprise class resilient solution. It includes high performance. So from an application perspective like storage is indistinguishable from using local flash. And of course cost efficiency. By separating storage and compute we get much better utilization of the storage as compared to using local flash. And we have data reduction capabilities. So for example, the within provisioning, the storage that is actually being used up from in terms of the raw capacity. Can be scaled automatically as you run out of raw capacity. You don't have to take provision. You can think, provision and only pay for storage that you're actually using. So there are efficiencies with data reduction, with separation of storage and compute that ultimately translate to much better cost efficiency for high performance storage. Right. Right. And I think people tend to think of, you know, I go to the cloud, I just check out a volume for each one of my workloads out of the native storage. But you're not really sharing at that point. And you're not. You don't have this same concept of efficiency you had in the data center with the sand that's delivering data to a whole bunch of different complex workloads. And now you can do that with light bits. Yes, exactly. So I'm glad you brought that up, because what we provide is a storage cluster. You can start very small with just three small nodes, and then linearly scale your performance and capacity by adding nodes with a storage cluster. The customer can use that available storage capacity and performance any way they like. It's not about assigning volumes to individual workloads and then managing those resources individually by for each of those workloads, we can manage in a unified way, bulk volume management and sharing of that available capacity and performance across many different applications, across even many different tenants, if that's what the customer wants. So it becomes a much more efficient solution, both in terms of the CapEx and the OpEx. And on top of that, we're able to provide the resiliency and the performance of the customer needs. Yeah, you mentioned resilience a couple of times. So we can do understand some synchronous replication to nearby availability zones and set up that, that kind of business continuity stuff that we expect to be able to do on premise. Exactly. So on Azure, for example, we can do three zone synchronous replication. That means if an entire availability zone goes down, the data is protected and the service is uninterrupted. Right. That's the kind of resilience you need for, you know, enterprise class, mission critical workloads. All right. And tell me about this this idea of performance. You keep talking about being highly performant. Is it very fast? Is this something that that, you know, people would just say, hey, this is really screaming now if I use light bits? Screaming and want to be clear, it's not just about, you know, selling a Ferrari here, which has very high performance. It's about doing it cost efficiently. So if your application needs, let's say, up to a million IOPs per volume, then with Littlebits you can get there. And with Littlebits you can get there with submillisecond tail latency. That's something that no other storage solution can do on Azure today. So scaling up, you know, for a storage cluster to millions of IOPs while maintaining submillisecond latency and, you know, doing it cost efficiently, we we have shown to our customers that we can achieve better performance than any other alternative that they're looking at on Azure at half the cost. So the cost efficiency, if you need high performance, we got the most cost efficient solution on Azure today. How difficult is it to manage and are there options for that. Do you offer this as a or have partners. How could I get like bits running for myself. Right. So on Azure we have a managed application marketplace offering, which means that the way we get deployed is we're running on top of Azure VMs, and we're pulling all that NVMe storage that's inside of those Azure VMs into a storage cluster. And then we're presenting that storage cluster to all the applications that the customer has that wants to use that storage. It can create as many volumes as they like and use that available capacity and performance. Now once they create the storage cluster, they can either continue to manage that on their own or they can have light bits manage it for them. So through the managed application model that Azure has, the customer can define the resource group and the light bits is operating and managing the storage cluster, which means that we do all the 24/7 monitoring. We provide H.R with SLAs, we do software upgrades. We expand the cluster when needed. We address Autohealing when a node fails because in the cloud eventually everything will fail. And make sure that the service is uninterrupted for the customer. So we are operating the storage cluster. There's two options then self-managed by the customer or lipids management of the storage cluster and doing that inside of the customer's own tenancy. So the data still resides within the customer subscription. So storage is a service but done in a clever way using the Azure Managed Managed service managed Application facility. This is pretty cool on there. Anything else you'd like to add or recommend to folks? If they're if they're trying to think in this direction about migrating to the cloud. Yeah. So bottom line is if the customer needs high performance storage and they want predictable low cost and they want zone resiliency. Then they should take a look at light bits. I think we have a great solution for for those performance sensitive workloads, and I invite customers to come to light bits Labs.com and ask for a demo or a free trial. All right. Because that would be my last sort of wrap up here is do that. So folks, you heard that. You know, check out the website. Check out the demo. Check out the trial. You can get block storage on several clouds now, including Azure as the most recent one, and move your sensitive and critical mission critical databases, those heavy things that couldn't be moved before into the cloud, and maybe finish some of those cloud migration initiatives you started years ago. Take care guys. Bye. Thank you. Can you.
