Storage not servers sees innovation

Server makers could be in trouble -- piling more woe onto HP's plate, perhaps, a company that underwent a firestorm of negative publicity following its recent dismissal of a CEO after just 11 months in office. But I'm not about to pour more fuel onto that fire. Rather, here are some thoughts on the future of servers and storage.

Research firm IDC has pronounced that server sales are set to fall over the next year, following poor economic news and a growth in the numbers of small to medium sized businesses moving their systems into the cloud. And server vendors are not beating a path to my door with announcements of new technology -- but storage vendors are.

Just in the last ten days, I've spoken to (among others) Zenprise, vendor of end-user storage security software; Red Lambda, vendor of a high-end grid-based security system; AppFog, developer of cloud platform services; Caringo, pioneer of object-oriented storage; and Vision Solutions which, after its acquisition of Double Take, is now in the disaster recovery, server mirroring business.

All had things to say around cloud and especially storage, where growth continues seemingly unabated. We're storing endless amounts of stuff. Big data equals big problems in the eyes of many organisations and shoulders the blame for a situation where backups take forever, where it's almost impossible to predict and cater for demand without busting the budget wide open, and re-organising the storage as a result of growth seems to be a regular not an occasional task.

Going to the cloud could be an answer in some cases, although big storage connected over the internet isn't advisable unless the whole computing system is close to the storage. The good news is that all the vendors I've spoken to offer something for cloud providers.

For example, object storage allows cloud vendors to organise their storage systems for redundancy and performance when billions of files are involved. Although most storage vendors mainstream on standard hierarchical file systems, object storage is likely to eat into that as users approach the limits of what traditional file systems can do. Mainstream vendors such as NetApp are starting to pay more attention to this market too.

Now, the growth of virtualisation implies an increasing separation between the server's duties and those involved in managing storage, object storage even more so since most of the implementations I've looked at involved a cluster of machines dedicated to the task.

Conceptually, object storage makes sense. Servers are getting bigger and, as virtualisation hosts, are addressing more storage. Right now, virtual servers inside that host use VMDK (or equivalent) files for their storage, files that are large and and take a lot of management when it comes to the protection of that data. What's needed is a highly robust storage system with a single global namespace, which can help reduce the management overhead as it scales. Object storage does that.

Best leave servers to do what they're best at - not managing storage.