In this episode I’m with Avinash Lakshman (CEO and Cofounder of Hedvig), and we talk about:
– the company
– architecture of the product
– use cases
Full transcript of this episode
Enrico: Hi everybody, and welcome to a new episode of Juku.beats. Today, I’m here with Avinash Lakshman, CEO and Co-Founder of Hedvig. Hi Avinash.
Avinash: Hey, how are you doing Enrico?
Enrico: Yeah, very good. Hedvig is a very new company, and I would like to know more about the company and what you’re doing.
Avinash: Absolutely. Hedvig as a company was founded in June of 2012, and we are primarily trying to deliver software-defined storage solutions to the market. My background prior to founding Hedvig was that I was at Amazon for three years where I was one of the co-inventors of Amazon Dynamo and then spend four years at Facebook where Apache Cassandra was my brainchild. These are the two big systems that I had delivered prior to founding Hedvig.
Enrico: As far as I know, your architecture is quite complex, but in the same time is very elegant, because you have this frontend modules and backend based on Cassandra. Can you explain to the listeners how it works and all the companies, how can they be deployed together?
Avinash: Absolutely. First of all, it would be extremely unfair to compare whatever we have built in over here as an extension of Cassandra, because it is not. Cassandra is a database and the Hedvig storage platform is anything but. The main aim of the Hedvig storage platform is to bring infrastructural capability that is locked inside companies like Amazon, Google, and Facebook to the enterprise with a great emphasis on flexibility, scalability without sacrificing enterprise-grade features, right.
If you look at the platform as a whole, it’s clearly separated into two different tiers. One which we call the Hedvig storage proxy. The job of the storage proxy is to masquerade as an iSCSI Target or as an NFS server. This was primarily to capture traditional storage protocols and then convert them into what we define as a Hedvig storage protocol. The cluster is broken down again into two logical tiers versus a clear demarcation between metadata and data. There is no such thing in a Cassandra system, so the system … Well, there were lot of lessons learned from my experience having built large scale distributed systems in the past, but it has nothing to do with the systems that I’ve built in the past.
Enrico: Well, it’s quite interesting, also because you’re promoting the system for a lot of different use cases, from virtualization to Big Data and Private Cloud. How does it work? How can it be possible for such a kind of system to do all these things? Where is the magic?
Avinash: It’s very, very loaded question. It’s going to be very hard for me to kind of articulate the answer in a single sentence. The only way I could probably do it in a single sentence is “try it and see it,” and that’s the only way to do it. If you look at the simplicity that we bring to the table, it’s pretty mind boggling, I think for the first time in the industry based on my knowledge, and I’m willing to stand corrected if you can show me other examples. One could actually provision storage assets from your iPhone or your favorite mobile device, right, that’s what we promised to bring to the table, and I guess we are delivering on that promise.
Right upfront, I’ve made a very conscientious choice to go very broad based from a use case perspective, because we didn’t want to be perceived as a niche player in any way or form. We have actually deployed the system catering to a wide variety of use cases. We have deployments in Big Data environments. We have deployments where we serve as infrastructure as a service from a storage standpoint for any hypervisor environment in particular VMware. We have deployments where we do serve as these storage infrastructure of choice for OpenStack environment. We have deployments where we do be a backup target so to speak where people wanted us to be replacements for their existing backup solutions. It is pretty broad based, and it’s been working out really well for us.
Now, if you’re going to … You can push the arm little further and ask me, can you be a replacement for the really high-end solutions where latency is absolutely critical? The truthful answer for that is I don’t know, but that’s the top pyramid of the market, and I don’t even believe that we would want to compete in that market at this point.
Enrico: Thank you very much. I have just a one more question.
Avinash: Please.
Enrico: How can we find you on Twitter or on the web?
Avinash: Yeah, if you go to hedviginc.com, I think our official website is very content rich. There’s a lot of customer testimonials. There’s a lot of information available about how the product works including a live demo of the product is available, and I think that should be a good segue to know what we’re actually doing.
Enrico: Thank you.
Avinash: It’s all right.