There. I've done it; shamelessly used sex to get you to read a column about storage. And I'm not proud of myself. But I'm not the only one who feels a need to make this subject more, um, exciting. While at Computerworld's Storage Networking World earlier this month, I was struck by how dowdy and unattractive storage networking seems at first blush - and by how exciting it really might be.
The basic story about disk drives and tape storage is that you can always buy more for less. That's a technical marvel but not hot news. The news, such as it is, is that storage has usually been "direct-attached" to a workstation or a server. Storage networking, the new trend, puts a bunch of disk and tape drives on a network so they can talk to one another. Breathtaking, eh?
As if that weren't bad enough, the industry is stuck in the same types of arcane standards wars and technical arguments that hit the LAN industry in the late 1980s and the e-mail industry in the early '90s. (How many of you know whether it's good or bad to mask a LUN? Or when to choose an arbitrated loop vs. a switched fabric?)But once vendors settled on standards and users learned how the technology could be put to work, LANs and e-mail became vital. I think networked storage, too, is going to be hot stuff within two years. Here's why:
It ain't the storage; it's the data. Storage, of course, exists only to store infrastructure software (operating systems, middleware, directories), business applications or data.
The core issue facing your organisations today is speed: That means fast, secure access to customer data, product data, pricing data, sales data, from both inside and outside your company. A scalable, secure storage management architecture won't guarantee success, but the lack of one could sink you.
Disks are cheap; people are expensive. The fact that storage hardware is so cheap is what makes it so dangerous, because managing the data can cost 10 times more per megabyte than the storage hardware. Networked storage is already saving some pioneers money. Jerry Lynch, director of operations at the Online Computer Library Centre, says he's seen his mainframe data grow 900% while his costs have risen only 42%, thanks in part to a storage-area network (SAN).
The data-storage problem will only get bigger. And it's not just e-commerce apps that are to blame. Hugh Ujhazy, director of strategic development at Ericsson IT Services, was among several customers who told of having to maintain several versions of huge ERP applications - one for production, one for development and one for test. Since moving from direct-attached storage to a SAN, he says he's cut his cost-per-megabyte by one-third - and he doesn't expect to add management staff even as he doubles the data he has under storage.
The magic is in the network. Just as vendors such as Cisco Systems built hardware and software to make systems networking possible, vendors such as Gadzoox Networks and HighGround Systems are developing hubs, switches and software for network storage devices.
The capabilities they're aiming at include highly granular security, where a certain class of users would be directed only to specific storage devices or quality-of-service offerings where the most critical transactions get routed to the fastest storage. All of which can help you adapt more quickly to the networked economy. Interested yet?
And the real magic is yet to come. Troika Networks and Hitachi Data Systems demoed a Fibre Channel SAN, which let them take almost all the storage-management load off the network servers, leaving more horsepower for the data mining app. The vendors claimed that they were able to scan a massively parallel cluster of DB2 servers on Windows NT at up to 1 million rows per second - 10 to 20 times what was possible with more conventional architectures. The Troika controller was able to detect, packet by packet, whether data was being transmitted using the SCSI or Virtual Interface protocols, routing it to the appropriate destination. That's the kind of selective service most mainstream corporate data networks don't yet offer.
OK, maybe the caterpillar hasn't turned into a butterfly yet. The vendors and industry groups such as the Storage Networking Industry Association may never solve a raft of technical, political and interoperability problems. IT managers may also balk at managing storage as a separate hunk of their infrastructure, along with their current network and application management tasks.
But the big storage networking vendors (not to mention some intriguing start-ups) are looking to hunt down and kill some really big business problems. Ignore them at your peril.