New tool for in-depth SAN management

If you're involved in managing storage in a datacentre, then you know that managing it is not always the easiest task. Trouble is, there's lots of companies out there now that promise to fix the problem.

When virtualisation started, the fact that it offered reduced costs and fewer servers to maintain and manage was enough. Now that virtualisation is the reality in most facilities, and that vMotion means those VMs are shuttling around the place like a tube train, the much-vaunted ecosystem of vendors that VMware made a big effort over the last ten years to attract has blossomed. You can't move for vendors flogging stuff that works with VMware.

But some are doing better than others. When I report that one of them, a SAN monitoring company called Virtual Instruments, claims to have increased its employee numbers from 60 15 months ago to around 140 now, you'd have to presume it's doing something that systems managers like, and that they like it enough to buy the company's products.

The problem VI aims to solve is not new: each vendor of hardware and software tends to have a siloed view of the world. It's natural enough, I suppose, to spend your development budget building software to manage your own kit but no-one else's. That doesn't help users in the real world, who need a much more holistic view that isn't fenced off at vendor boundaries.

So VI has just launched a new generation of its VirtualWisdom product suite, a combination of hardware and software. The software runs on Windows Server and acquires data from software and hardware probes. The hardware probe sits alongside a traffic access point, which splits the optical Fibre Channel signal and diverts some of it to an analyser. You connect this to a SAN probe device that converts the optical signals into electrical ones which can then be analysed to identify and resolve problems.

Although it's a bit scary inserting hardware into the optical path, with a redundant SAN architecture you can probably get away with it with minimal or even zero downtime. VI's system then allows you to see latency values, transmission errors, SCSI reservation conflicts -- for example highlighting slow vMotion transfers - and so on, all in real time.

According to company CEO John Thompson: "the system will show where latency is really happening per server per transaction, not an average over the last 20 minutes. People running mission critical applications such as order processing or financial transactions really need this."

VI claims no-one else is approaching the problem in so in-depth a manner, and that its sales figures show it's on the right track. It also reckons there's a FCoE product on the way, out in about six to nine months.

The company's average transaction size is about half a million US dollars, though a starter kit can cost as little as $50-100k.

Of course, the best time to do all this inserting into the optical path is before you start to have problems, so that you have a clear baseline of problem-free data. And of course, the only time you'll start to seriously think about and get budget for such an implementation is during or after a big SAN outage....