View from the Hill

The Next Decade in Storage

ACSL, publisher of STORAGEsearch celebrates its first 10 years of computer directory publishing in December. As editor, I thought it would be interesting to speculate about what major changes the next 5 to 10 years might bring. We'll be returning to these and other subjects in much more detail in future articles.

Who will dominate the storage market?

Is the storage market going to be dominated by a single supplier? in the same way that the IP switch market is dominated by Cisco, and the Unix market is dominated by Sun?

If you'd asked that question at the beginning of 2000 the bets could have gone either way, and the answer might have been EMC. But in 2001 we've seen the start of some irreversible trends which will shape the market of the future. EMC lost 9 points of market share in the external RAID market this year, and the biggest gainer was that category (which includes hundreds of RAID companies) and which market researchers lump together as "others".

It's clear that even at this early stage of the new storage market that users regard network storage as a commodity, and don't see why they should pay a premium price to anyone for a box of disks with some network ports. Storage will end up looking much more like the PC market, in which there are thousands of manufacturers. It will be difficult for a single storage company to capture even as much as 10% of a market which will be worth hundreds of billions of dollars. No single company will dominate the market.

The end of operating systems?

The increasing use of data network technologies like XML and storage virtualization software in new business applications software will reduce the role of the operating system in the computer market from the primary buying criterion, which it is today, down to a secondary minor role. When you can do pretty much the same things with your data whether your OS was written in Seattle or California, or as an Open Source project, the OS is going to become as irrelevant to most users as the source of their gas or electricity is today. It's only the computer appliance manufacturers who will interface at this level.

Of course, the desktop appliance, which we nowadays call the PC, and the notebooks etc will continue to be mostly Microsoft Windows based products, but as long as they get shipped with all the connectivity they need, users won't really care what the differences are in the internal versions, because, as now, they'll be driving these things from a browser front end. In the long term, we may even see the disappearance of the reset button, but remember I'm talking about a 5 year timeframe here, so the mechanical switch manufacturers don't have to start panicking just yet.

The end of tape backup?

The tape library occupies the same slot in the IT datacenter arsenal today that the ironclad Dreadnoughts did in the Europe of the early 1900's. They're expensive to buy, include a lot of metal, and are seemingly invincible.

Owning more Dreadnoughts became an obsession to navy planners in the UK and Germany in the years leading up to World War I, because they demonstrated superpower status. In a similar way, owning a fleet of tape libraries indicates to the outside world that your company is a massive data owner, such as a media company, a bank, a telco or other corporation which is on the same scale datacenterwise as a government department. So you may get a bit twitchy when someone predicts that you're going to pull the plug on all that investment, especially when most of it hasn't even been installed yet, and is waiting for the next budget period to kick in. Well, remember, I'm not talking short term here, but here are my reasons.

Tape was a good idea as a backup and recovery technology in a disconnected world, when disk drives were expensive, and data security depended on being able to carry your data into a car for off-site backup via sneakernet. Although the density of tape backup has increased, so too has the volume of data which people want to store.

Data weighs a lot, and the average person would not feel comfortable carrying a terabyte of storage for very long. Unfortunately the terabytes are are growing like Topsey. Tape libraries solve today's problem of backing up data networks, but no-one suggests that you're going to unplug your tape library, lift it up using a fork lift and drive it to an off-site location as your secure backup. Get real. The way that tape libraries manage the off-site backup problem nowadays, is they use IP based data replication software to back themselves up onto other tape llibraries somewhere else...

...And that is exactly my point. If you aren't going to pick up the whole damn thing and move it, then there is no particular advantage in using a tape cartridge as the medium for the data replication. It could be any convenient, reliable technology which stores data, such as a RAID system using hard drives or an optical based juke box. So one of the historic arguments for using tape media has already been junked. The internet doesn't care what shape or size the media is at the other end.

I think tape will put up a fierce rear guard action, and remain a factor in the data recovery market for many years, but its days are numbered. From now it will only lose market share, maybe just a few points each year, but the writing is on the wall.

...I look forward to reporting on all these changes and more, in our 2nd decade as a computer directory publisher which started this month.