Infinio, interview with CEO Arun Agarwal
Date: 23 November 2013
Infinio CrunchBase profile
Infinio
Centralized storage often represents the performance bottleneck of the datacentre stack. The storage array sits far away from the application and has to deal with changing I/O from multiple applications. As a result, latency to a busy storage array is high and performance can be unpredictable.
Moving storage closer to the application is a great way to solve the problem. Server side flash is close to the application and can act as a buffer to steady reads and writes to the centralized storage array.
A barrier to overcome with most solutions, is that customers are required to order hardware and invest in considerable work to perform the installation.
Recognizing these barriers, Infinio differentiates itself by offering a virtual appliance that uses existing server memory to cache reads. To sidestep any remaining organizational barriers, the product can be downloaded and fully tested for 30 days.
I Interviewed Arun Agarwal, CEO of Infinio to find out more about the company and the product.
After you finished Harvard and MIT, what steps led you starting Infinio in 2012?
I actually built my career in the Boston area. The first start-up I was part of was Endeca Technologies, that was bought by Oracle for 1 billion dollars. I kind of have been a start-up guy ever since.
I entered VC for a little bit, I worked at Lightspeed Venture Partners and Bessemer Venture Partners. I really wanted to go back building a business and that’s how I ended up working with my co-founder Vishal Misra. Through our connections at Columbia and my connections in Venture Capital we actually saw an opportunity to build a company.
Two of the VCs I worked for are actually backing Infinio and I have been previously involved with Highland Capital as well, which is our third investor.
These connections I built up in the past all lead to starting Infinio. The best way I can characterise it is that you have to create serendipity, keep your ear to the ground and be opportunistic.
It’s a little bit of art, little bit of science and a little bit of luck
If I were to describe what Infinio does in two words, it would be read caching. Right?
That is the technical story. What I would say is that Infinio focuses on two things; on-time, on-budget.
The question that should be asked is: How can I deal with the problems I have? What we are trying to sell is the simplest, easiest and most cost-effective solution to a very painful problem in the datacentre.
We are trying to sell to the person and not just solve a datacentre problem. I think that is being highly overlooked in our in our industry. I think some people are really getting caught up on the speeds and feeds, it’s all about the microseconds, the milliseconds.
I am not trying to trivialize the importance of that, but at the end of the day there are people that are running datacenters and we should be focused on how to help them out.The goal should be to keep the end users happy and keep costs in control.
Infinio is virtual appliance. Added overhead and latency are associated with that choice. Why didn’t you build a kernel module?
That is a good question, let me sit back and talk you through what the goal of Infinio’ s product is.
Infinio is less about making storage run much, much faster. We are more about how do we keep storage boxes out of trouble.
We say that the vast majority of applications out there are fine with millisecond latency. Where these applications get in trouble, is when these storage arrays get overloaded and applications are halted because of latency spikes.
Infinio is trying to essentially offload storage to extend its life dramatically and to avoid having to throw hardware at the problem. Infinio is more of a performance utilization play than a performance speedup play.
We focused on: How do we make a solution that is extremely easy to download, try and install. A Virtual appliance is the best choice for this. If we would have been after speed, the lowest latency then we would have made a difference choice.
Now of course we are dealing with storage performance, but the differentiated position we are taking in the market is not about going faster. It is about buying less hardware. Our tagline is buy performance not disk. Personally I think that is a better business.
Imagine a setup with 3 hosts running VMware, fabric switches and an EMC SAN. What happens if the Infinio virtual appliance is not available/crashes?
So for every system running multiple hosts, there is an Infinio virtual appliance running on every host plus a management virtual machine. When the management VA detects a failing Infinio VA, the first thing we do is unplug it and boot the VA back up again. We simply get out of the way , so that in the worst case the setup performs like it would without Infinio.
What are the main use cases your customers use Infinio for?
We are actually seeing a wide variety of use cases. Any place you can reduce hardware spend it is worthwhile to test our product. We are seeing good results across a wide variety of workloads.
If you were to consider competitors, what names come up?
Names that come up when people compare us are PernixData and Proximal Data. We actually tied with PernixData as a finalist for the best new technology at VMworld 2013.
I believe that, in the case of PernixData, if customers are seriously looking at both solutions, then one of us is in the wrong place. The solutions are just too different.
How large is your current team and how are things going?
We are currently around 30 people and we raised around 12 million dollars from top notch VCs.
Everybody is really psyched, things are going really well.
What are the next big steps for Infinio on go-to-market and product development?
There are a few major steps coming up. Currently we are NFS-only and we are soon adding block support to our technology. We want to serve as many customers as we can.
Second is that we want to work toward 100% self-service. Our product is already very easy to install, but we are still working with customers quite closely. We want it to be completely self-service, where customers install from a portal and we are 100% hands-off.
With that last goal reached, it really all comes together. We have a product that is so easy to install, that it enables us to offer it in a self-service way. That makes our internal sales process much cheaper, allowing us to set the price lower and offer more value to our customer.
What are your goals for the coming years?
The idea is to build a great storage company. We have everything we need to do that. We have the right team, the edge, the positioning and the technology. So we are going to build an interesting business here. We will expand our platform support, add features and just expand our market share.
The higher level vision I have for the company, is raising the bar for easier to use, easier to try and easier to buy. Those are all things that I believe Infinio can apply to more than just this one problem.
Background and Links
Infinio website
Great introductory article by Duncan Epping with links to many Tech Field Day video’s
Pernix Data article; Interview with CEO Poojan Kumar
In depth article on Virtual Appliance vs. Kernel module by Frank Denneman
Crunchbase profile of Infinio
Geef een reactie