Server SAN software upstart Maxta gets Intel's cash inside

Server SAN software startup Maxta has gained $25m in a B-round of funding led by Intel's investment arm and Tenaya Capital just as the software-defined and converged storage trends are taking off. The upstart swallowed $10m in an A-round in October last year. With converged compute/storage hardware-software competitors …

[Jim 59]

What is it

The technology is being hyped as if it contained new ideas or immediate benefits. But neither is immediately apparent. The concepts - VM appliances, local disk, clusters - are all pretty ancient. The solution - essentially clustered NAS - is complex, and potential benefits in performance, price, reliability are not obvious. Price is a possible advantage but may be undermined by clusters being labour intensive to maintain.

El Reg has not yet given a sober review of this tech, but commentards can find a reasonably helpful analysis here, complete with a handy guide to the Gartner Hype Cycle.

[Anonymous Coward]

Re: What is it

I agree that it is not new. The reliability and performance benefits over traditional SAN are pretty obvious though. The cost should be less, if we are just talking about a bunch of inexpensive x86 servers with a clustering platform, but I suppose they could make it as costly as they want to make it with software add ons. If SAN server means Google's architecture, that is inexpensive... it is complex though. That is why the route most people will likely take will just be to let Google, Amazon, etc manage their NAS for them and worry about the complexities of clustering. There is no reason the average company would, or should, want to take all of this on when the cloud providers are willing to do it for you for extremely low costs.

[K]

Re: What is it

"There is no reason the average company would, or should, want to take all of this on when the cloud providers are willing to do it for you for extremely low costs."

Gave me a smile for a Friday afternoon. Let me give you a few reasons -

1) Cost - If you had a multi-site company, using "Cloud" based storage as "NAS" the costs for the requests would be massive

2) Bandwidth - Most companies don't have the necessary bandwidth to facilitate the amount of data, also downloading a 20Mb file accross a 100Mb leased line is dam slower than loading it across a local 1Gb network (Hence frustrating for staff and a productivity killer!)

3) Latency - The average latency on a local network is < 1ms, with cloud it could 200-300ms (or higher, depending on server locality), Bear in mind loading a single file takes multiple requests, this could add several seconds to load times.

4) Resiliency - As an IT pro, your first priority is being guardian to the companies data and systems, this includes you taking responsibility when the cloud operator goes tits up!

5) Compliance and Security - Pretty much self explanatory. Do ignore that spotty junior admin at Acme Cloud Services perusing through your "secure" storage in search of some juicy photos!

Now if your still convinced cloud is the answer to all your storage woes, then step in my shop, I've got a product you might be interested in, Dehydrated water - its going to be the next big thing!

[Bruce Hockin]

Re: What is it

SeverSAN has a big future of performance optimised storage. It's not the only route, there are plenty of other choices but the performance vs cost arguments are very compelling. It won't be for everyone, and for others there are all flash arrays, hybrid arrays etc. There's a home for all of them. And yes, the basis of the concept has been around for a while, but a number of things had to change in order to facilitate the growth of this technology - the key one being the accelerated pace of solid state memory development, both from a technical and cost standpoint. Drawing parallel's, you need look no further than object storage. It's been around for ages, yet why didn't it take off before? Customer need. Problem is object storage is now not a technology looking for a problem. Explosive growth of file data is the problem, and object is now the only technology which can scale economically to support it.