scale computing

Over the last couple of weeks I have been brought into the fold and made very enthusiastic of this new product from Scale Computing, our American cousins have brought out a product that is very different, so different to the point where the analysts have had to invent a term for it called “hyper-converged computing.” Whilst I don’t like the American need to pollute our language with words made up on the spot, I do like a lot of their technology (and cartoons, and food, and New York).

Unleashed IT are one of Scale Computing’s UK partners and we’ve been on-board from the early days of their entry into the UK, having a couple of interesting clients already on their technology just for storage. Scale’s storage products were simplicity in themselves, but before I explain this, I should probably explain where the company has come from as I learnt from Jason Collier their Chief Technology Officer at IP Expo this week…

A lot of the Scale execs are seasoned start-up technology gurus and have a number of success stories under their belt. One of the projects that Jason and team were working on a number of years ago was to build a trading platform that had supercomputing speed for a lot less of the cost, so forgive me if I get my figures slightly wrong here, but essentially they built a super computer for $60k out of standard server components – parallel computing and all that malarkey. So what do you do when you build a cheap supercomputer? Well, you need storage, so they went out to the big vendors such as EMC, NetApp and IMB to find that the cheapest they were looking at was around $200k to have something that could perform at the speeds of the supercomputer – IOPS as the storage trade call it.

Hence Scale Computing was born, an affordable storage system that has incredible performance to keep up with the needs of just about anything. What they then went on to build was simplicity in itself. Using off the shelf servers they developed software that was based upon IBM’s supercomputing file system, which spreads data across a storage cluster, making the thing fast and resilient. One of the anecdotes that I heard was that IBM were even amazed at what Scale had accomplished automating a lot of the procedures that they have to carry out on their file system manually. Everything within a Scale box is being run via a decision engine that is constantly looking at the state of the software and hardware and trying constantly to get it into a stable position.

So to start a Scale Computing cluster, you need a reasonably quick network switch as this is where the communications between the boxes is done. Then as a minimum you need three nodes (servers to you and me) in a cluster, you can then add more nodes to your cluster as and when you need, giving you more storage. What the system does is decide where to position data and takes care of that for you. Even in the minimum cluster size, you can lose an entire node and the system will operate, you can lose several disks in a node and again the cluster will operate. Performance will be affected but your data is safe.

With Scale Computing HC3, what they’ve done is taken this concept further. You essentially have three servers in your bare bones cluster, all of which have processing power and memory that is entirely dedicated to running storage. They have taken a decision to now utilise this to provide virtualisation capability. So you have the storage, memory and processing power in one place, with one control panel. I’ve not seen a system easier to build a virtual machine and deploy it and the great thing is, because it’s built on the back of this decision engine – you’ve got no vTax (I’m told that’s the term) for System Center or vCenter. The system uses the KVM hypervisor from the Linux stable, meaning costs are very low and the technology to move virtual machines on an ad-hoc basis when there is a node failure is all done through the technology Scale have developed.

I as a techy was very dubious about not using a leading hypervisor such as HyperV or ESX/ESXi – however, I did eventually calm down knowing that the virtual machines from these could be run on the Scale box, so if you trial the kit you can move it around as necessary. In general on the back of the trials we haven’t seen any of the Scale kit returned. It is very simple, easy to manage and pretty much does what it says it’s going to.

But, this is the amazing bit, you’re getting the technology that the big boys have at a very reduced cost with the added advantage of it all being managed in one place – people have tried to hook their storage into the VMware vCenter control panels, likewise for HyperV and System Centre, but I don’t personally think many are that effective. Scale has all this in one place and in one appliance (well many appliances that act as one, but you know what I mean). The analysis’s are saying it’s a datacentre in a box – their demo kit as pictured in this blog (which coincidently, if you need one, we can provide!) shows just that, a small datacentre in one box. A datacentre in a box – ideal for business continuity plans!

The demo unit pictured above is called Skywalker – imagine it on a ramp with a break off cable so that when the fire alarm goes off it rolls out of the door. Okay I’m kidding about that bit. However, for medium business this is an ideal product to run infrastructure on, for larger companies it is an affordable piece of kit to hold in a contingency store.

You can obviously tell that I’m enthusiastic, I’ve stuck with Scale for a couple of years now, I haven’t been all loved-up to be honest. I felt I didn’t know where they were going, that their dipping their toe into the UK market wasn’t done properly and a lot of it was put on us to push their products. Customers all thought the kit was great but weren’t that happy about taking a risk on a company they’d never heard of. Well HC3 is the where the company was going and was always their intention and the development goals of where this is going will also blow customers away. Scale has significantly increased their staff and resources in the UK which has made me feel very confident.

Don’t get me wrong, I am a Dell and Microsoft man (yes we do sell other things before you ask), but me personally, I love Dell hardware and Microsoft software. I am a fan I stick with them through the good and bad decisions, but it is incredibly refreshing to be dealing with a company that is new, that goes back to those ideals of when Dell and Microsoft were young and on the up. It’s nice to deal with a support call centre that’s in Indiana and not India! I have actually reconciled that it is time to try something a bit different.

I also have a lot of faith having spent a bit more time with CTO, Jason Collier over the last couple of weeks at various events including IPExpo, anyone who can wear 37 badges on a pass lanyard and be proud of his 37 pieces of flair, in my book is as geeky as me. Office Space is one of my favourite films and that of most geeks so if you need to get that joke, then watch that film.

We’d very much like to talk to anyone interested in Scale Computing directly, hook you up with our existing customers so you can hear directly how good it really is.