Getting ready for EMC World
Steve Cassidy is getting very excited about storage, more specifically EMC’s VSPEX architecture.
I am by now, used to people getting puzzled by my obsessions. I am not referring to the truly eccentric ones, like knowing how to clean ball mice or how to take apart an early Mac Laptop (yes, it was been one of those bank holidays, thanks for noticing). I am thinking very specifically here about storage. And so should you, because EMC World is almost upon us. Next week, in fact. (Ed: And IT Pro's news editor Caroline Donnelly will be there reporting on all the news and views.)
Why would a storage company feel it was in the business of setting standards for deploying virtualisation?
OK, in part this is a snobbery test. To know who EMC are, you need to inhabit a particular part of our IT ecosystem, or be in the habit of passing through Heathrow's Terminal 5 fairly often, or have been part of a network deployment so massive that the contents are very likely to be secret.
Outside of those various cliques, EMC is not likely to be a very big name for you, though there's a very simple justification for that to change. Way back in the late noughties, just when my PC Pro colleague Jon Honeyball and I were beginning to sniff around virtualisation and see the lay of the land, the only place to go for a conversation was VMware. At that point the virtualisation giant had two equally distant and improbable fairy godfathers. One was Cisco and the other was EMC.
Cisco is a little bit more widely understood, and in many circles treated with some caution. It certainly makes at least a bit of sense that a massive networking giant would want to invest in virtualisation.
It also makes sense for a vendor of very heavy duty storage to be investing there, too. So much so that it wasn't a big secret that VMware wanted to get out from under the thumb. This is likely because a lot of storage architecture work at the bigger scales doesn't look or feel like it does down in the world we live in. The world, that is, with single servers doing single jobs even now or of clients who can't understand why virtualisation only marks the start of a new type of spending pattern, rather than the end of spending altogether.
This goes some way to explaining the emergence of VSPEX from EMC. Why would a storage company feel it was in the business of setting standards for deploying virtualisation? Because it's nothing like as easy as it seems, that's why.
EMC is about to host its own convention the aforementioned EMC World - out in Vegas (I guess that explains all the adverts in Heathrow T5) and even the most casual perusal of the surrounding releases, announcements and stall inhabitants will show you that there's a hell of a lot of people out there worried about management, monitoring and analysis.
Some are also misinformed, falling for the commonly thought belief that virtualisation is the source of the "more space, more space" problem. But that isn't the only cause. Even if all you do is stick to the letter of data retention laws because you do business with Americans, you will still end up with a storage problem. Basic data is bad enough, without virtualisation added to the mix...
Don't even go so far as big data that's a club for those who have the scars from two or three rounds in the ring with some petabytes. Well before you qualify from the school of hard knocks big data, you will discover just why there's a whole lot of companies who make a perfectly good living hanging out in the EMC customer base (and the others too), selling management and monitoring tools.
I don't know a single person, inside or outside commerce, the business, the tech journalism business, academic research not a single person - who has made their first try at a 10-terabyte storage resource actually work as intended.
The whole field is dotted with embarrassed first tries, NAS boxes locked away in cupboards, sudden emotional outbursts about iSCSI that you didn't see coming, desperately undersized budgets, the list goes on. Storage is a field with somewhat hidden non-linear traps in it, and not all of them are technical.
This is what VSPEX is potentially going to help with. Recently, I have been looking around a project that stores as much as possible of the media archive of the Montreux Jazz Festival on a big data storage array. This is an extreme case of the end result of the process of business analysis that has to be done and done well to make anybody's larger storage project actually work and keep working.
EPFL (AKA the University of Lausanne) was the deep thinker on the Montreux project and its key realisation was that the tricks that make corporates happy with big data de-duplication, thin provisioning and object storage were only partly applicable to the nature of the stuff it needed to serve up.
EMC's stratification of products is a step down that same road of appraisal. Most of the benefit happens on paper by which I mean that it's during the scribbling of shortlists and the construction of architecture budgets that this will ease the pain. And in the picking of technology sellers. Many firms pick a specialist vendor to implement this kind of system. Some even have the vendor do a reverse outsource, and plant a load of kit inside the buyer's computer rooms, which remains their property and responsibility.
The great intellectual challenge in that situation is whether said vendor can meet its performance targets and promises. People like me love the idea of an immensely complex vector space of multiply co-dependent you're bored already, right? With VSPEX, EMC has taken me and my sort further out of the specification loop.
Click here for coverage of last year's EMC World.