Jeda Networks, interview with Stuart Berman
Date: 5 Oktober 2013
Jeda Networks CrunchBase profile
As a part of a larger series, I am interviewing CEOs that lead companies in the field of Storage, Virtualization and Networking. What developments in the market did they see that inspired them to build their company? How will the market develop from here? These questions are on most of our minds, and I am keen on asking leaders in the industry for their views. This article is followed up by part 2 with background on Software Defined Storage Networking (SDSN)
Jeda Networks created a Software Defined Storage Network solution that reduces complexity in the connectivity between the compute layer and storage layer. They abstract al the complexity into software and then use regular Ethernet fabrics to build these virtualized storage area networks.
They are in a specific niche in which very few (if any) have dared venture. The inspiration for contacting Jeda Networks goes back to an article by Duncan Epping that I read a while back.
CEO of Jeda Networks Stuart Berman and I had a call last week in which he provided insight into the start of the company and his vision on the Networking and Storage market. Jeda Networks is Stuart’s fourth startup that started 4 years ago. The company was self-funded for 2 years and received external capital 2 years ago.
Those first 2 years must have been tough and exciting times
Well it is actually my fourth startup. Starting up new technology companies is becoming a sickness of mine. On the one hand I know the tremendous amount of hours that it takes to start-up. On the other the hand, since I’ve done it a few times before, you sort of know all the pieces. That tends to accelerate a lot of things that you would have not been able to accelerate prior.
I have been involved in four startups: Prodigy Communications, Bayfront Technologies, Arcxel Technologies and Jeda Networks. In all of the startups I did I started out as the CTO and developed the technological idea.
How is Jeda Networks doing now?
Currently we are at 17 employees and we have the software in Beta testing. Additionally we have a bunch of unannounced OEM partnerships that are testing our product right now. Our product is currently getting battle hardened, used in different environments. We are expecting to launch our product in the first half of next year.
When building enterprise class software that runs in a datacenter that can never go down, it’s always going to be a white knuckle ride. Basically It is an exciting time and we are all working hard.
Relatively speaking, software defined developments in networking seem behind on the Server and Storage market
We focused on the storage area network to virtualize. That has been one big piece of the IT infrastructure that has not been virtualized. You have storage arrays that have been virtualized for a long time, and there have been software defined networks for local compute-to-compute areas. But nobody has really applied the abstraction of virtualization to storage networks.
There have been 2 companies that have come close to applying virtualization to storage networks, Xsigo and Next IO come to mind
Well Xsigo was sort of virtualization in a box, Next IO was sort of virtualization on the server end. No one has taken on the storage area networks, which is a 2-3 billion market like we did. We abstract al the complexity into software and then use regular Ethernet fabrics to build these virtualized storage area networks. We’re compatible with all the storage developments out there and compatible with all the innovation in the servers too.
How do you see the future of storage in general and how does your product fit in the these future developments?
That is a good question because not enough people see their technology in the proper context. I see our technology in the context of the different types of IT infrastructures you see today;
1) Single purpose datacenters: Facebook, Google. They have driven the economics way down. Very low cost because everything is free, enormous scale, and the speed is really high. So that is one environment that needs a specific type of storage.
2) Cloud environments: Rackspace, Amazon and others. These environments run middle-of-the-road applications that don’t require a lot of performance in terms of IOPS or resilience. Examples are lower level web environments, email.
3) Enterprise environment: Where you need very high resilience, very high performance to run your Microsoft SQL, Oracle, SAP environments. These are applications for which people get fired if they fail.
4) High performance small-to-medium business environments; Ones that run one or a few applications that needs lots of performance. Applications like VDI or video-editing of 4k content.
If you look at all those environments, our technology is really focused first and foremost on Enterprise environments. Second we are also focused on the cloud, because you cannot put certain workloads on the cloud today. Workloads that require high performance and reliable access to storage. Today, for instance, you cannot put a SQL server application on Amazon. Our technology might migrate certain of these applications to these cloud vendors. That represents a huge opportunity for them.
So, when I talk about the future of storage I talk in context of all these areas. I believe that with our technology we can focus in on high performance block storage in several of those areas.
Do you recognise the trend toward decentralized / server-side storage and how does this impact your technology?
I heard the keynote of the CTO of Pernix Data at the storage developers conference in Santa Clara (SDC) a couple of weeks ago. He brings some good points. There are some difficulties to bring the SSDs to the server side, closer to the application. He showed data that shows how applications benefit from server side SSDs, and how their solution solves some of the difficulties.
I believe there will be applications that will benefit or require server-side storage. At the same time, I believe there will remain great benefit to having many drives in a single location. That way it is easier to build networks in enterprise traditional ways. Networks that have distributed storage find there is a lot of congestion when all these storage nodes start talking to each other. So I think that everything is going to grow, and certain infrastructures will be targeted by certain solutions. My vision of the cloud environments is a mix of server side SSD with nodes of many disks.
How do you see Hyperconverged technology like Nutanix and Simplivity are offering?
On a technical level, if you have a finite set of compute resources and they want to access a multi-terabyte dataset reliably and they need to do that quickly, then it will be hard for all those boxes to communicate with each other and manage that. Rather, having a rack of dense compute resources tied together with 40Gb or 100Gb network connectivity with terabytes of storage behind it can be managed, because it is designed to do it.
I believe Nutanix and Scale IO have their segment of the market. But I have sort of been through the contraction and expansion of the universe a few times in my career.
There is room for everybody in a growing market. At the end of the day what helps Jeda Networks is that the enterprise IT manager depends on reliable access to data and having these mission-critical applications stay up. He is slow to migrate off to new technology, we’re giving him the capability to grow his infrastructure to next generation applications.
In relation to hyperconverged technologies; we can actually duplicate the IT in a box in a bunch of racks. Within the space of a few racks we can take servers with lots of compute resources and map those with our technology to several nodes of storage, that you can configure in many ways. The all-in-one by comparison, has traditionally been restrictive in the requirements they could satisfy.
It is interesting to me how the future of the hyperconverged market will be. If it starts to take off, why wouldn’t Dell, IBM and HP make their own IT-in-a-box solution? It’s a rough and interesting market.
What has been the driving force behind starting Jeda? What drives you to get up in the morning?
It’s all about seeing a need and having enough knowledge in that domain to do something about it. I saw a need where storage network technology was getting too complex while the world was going towards very agile and scalable data infrastructures.
While I looked at that, at the same time many people were looking at Software Defined Networks, which is being applied to nearly everything today. Contrary to storage networks, I believe that with regard to Local Area Networks it will take a while for SDN to be adopted by the market. I believe adoption depends largely on the value proposition the technology represents. The ROI on SDN is not evident at all.
In contrast to SDN applied to Local Area Networks, we believe that with our Software Defined Storage Networking solution, we offer great ROI. We therefore have a market today, where we make Storage Networks easier and lower cost. In the future, we see an application to move workloads from enterprise to cloud. Both scenarios should mean good business for Jeda Networks.
I liked your reference to Marc Andreessens article. Besides recognizing such trends, however, the key point for me is to see the right trends and attach them to value propositions that you can offer your customers now.
Marc Andreessen is backing Convergent IO (now COHO Data), is he involved in Jeda too?
We have 2 investors, US Venture partners and Miramar venture partners + 2 unannounced private strategic investors. On future rounds I would definitely engage Andreessen and Horrowitz, because of their vision. Marc Andreessen and Peter Levine are amazing individuals so in the future we will definitely want to talk to.
Looking back on where you are now, who have helped you to get to where you are?
I attended UCLA and one of the professors, Leonard Kleinrock worked at the predecessors of the Internet, ARPANET. He was one of those brilliant people who took very complex material and made it very simple and interesting.
Then there were guys like Mario Gerla and Wesley Chu who were professors but also had lots of experience in the private industry in the communications area. That was very interesting to me, because it was the nascent era where they were connecting all these technologies. They have really inspired me and from there on it is has been a great ride.
This article is followed up by part 2 with background on Software Defined Storage Networking (SDSN)
Background and Links
Jeda Networks website
Duncan Eppings Article on SDSN
Marc Andreessens article
Geef een reactie