Address-based storage is the past: Introducing content-centric storage

We spoke to Arun Agarwal, co-founder and CEO of Infinio, about why a content-centric approach to storage is the most intelligent perspective from which to view the future of storage.

There’s a lot going on in the storage ecosystem. Nowadays you can’t trip without hearing about SSDs, and now virtual SAN is coming in and changing the way we store data. Things are changing very fast, and it’s not just about investment – it’s about companies getting to grips with the new technologies.

But any new storage system has to work with what’s already in the data centre if it’s going to be effective.

The big problem with storage today is that it’s address based, which means it’s held in certain locations. Whereas we look at data and think that it’s just data – it’s self-described. So today if you want to watch a YouTube video, you’ll go to YouTube – everyone will go to the same place on the Internet to see it. And that’s massively inefficient.

If we look at the world differently, then you can imagine doing things a different way. For instance, the first person could download the video, and then give it to everyone else. It’s an intelligent, content-based approach to storage. The whole world and the whole network can work much better if you look at storage from a content-based perspective. Then when you look at acceleration, in-stream de-duplication, security and encryption – all of these start to look a lot more intelligent once you look at them from a content-based perspective.

Content-centric is about looking at the data and not its location. It’s about looking at what kind of data it is, and therefore what’s the best pace to get it from – and it’s going to transform a lot of areas of the storage industry.

The main advantage of the content-centric approach is that it’s an incredibly efficient use of hardware. It saves huge amounts of server resources, and has a very low footprint in the data centre. Everyone looks at IOPS for dollars – they look for performance in a very singular way. I think the problem is that people get obsessed with single metrics – but the main goal is simply to make users happy. And if you can make the system work, the users are happy.

It’s like with the battle over smartphone cameras. Everyone’s talking about 10 megapixels, 20 megapixels – but no one is actually taking these pictures and blowing them up to the size of a wall. Past a certain point, the metrics start to become irrelevant.

The problem is, if you look at a private cloud, you have your servers and you have your storage – and then you hit these bottlenecks. An in the old world, the solution is to go out and get more storage. And you end up with all these half-full storage arrays that could hold more, but they can’t for performance reasons.

And there’s a lot of people doing a lot by way of acceleration, but what we’ve done is we’ve developed a purely software-based solution.

Now instead of getting the video straight from YouTube, we all get it from our neighbours, and it saves huge amounts of our cache space. When all you’re asking for is 8GB of RAM, you can be really non-invasive and slide right in. And that’s better than going out and investing in flash and installing SSDs across the whole data centre – which no one in IT wants to do.