Published: 09 Feb 2005
OK, don't put the magazine down, just glance over the top of it. Can you see your storage area network (SAN) or your big network-attached storage (NAS) boxes? Got your feet up on the desk? Maybe you're thinking, "Gee, after a few years of trench warfare to get this stuff to work, things are pretty stable. This year should be pleasingly calm. No change is good change."
Was that good? Now stop daydreaming--your work is just getting interesting. We're entering a new phase of storage networking. Fibre Channel (FC) SANs and high-end NAS are known entities, as well as relatively mature and stable technologies. Many of the bugs are out and prices continue to plummet. Maybe they'll never be true commodities, but their success has certainly inspired technologies that will be, such as iSCSI and SATA RAID.
In belated New Year's pundit fashion, my view of the future of networked storage is far from just a cheaper version of what we have today. We're on the eve of a second generation of storage networking.
The first generation was based on two simple ideas. Remove the 15-connector SCSI cable between a server and its RAID cabinet and replace it with an FC cable. Cut the cable in the middle, add a switch and multiple servers on the front side of the switch and, voila, you have a SAN. NAS was an even simpler idea: Take a network file and print server, lose the print part and make it large enough to be a viable network storage repository for many servers.
The second generation of storage networking is based on more complex notions, not just refinements of the first generation. You can already see some of the key concepts that will guide the second and possibly third generations of storage networking:
Content awareness. Newer storage will reduce the ways in which data is needlessly duplicated. That alone will probably be a double-digit savings in capacity at most companies. At the end of this line is bit-level, single-instance storage, although how many shops will go that far remains to be seen.
The triumph of the logical over the physical. With vendors sniping about whether to virtualize at the host, switch or array, it's easy to lose sight of the overall point: Eventually, you'll present storage as a purely logical entity, and your hardware and storage will manage the physical layer.
Location independence. The next frontier for storage networking is the wide area. If latency can be dealt with in a way that preserves data concurrency, then why wouldn't you centralize a host of data protection and availability functions?
Time, the fourth dimension. The graceful aging of data to appropriate storage devices remains a thorny issue. Simple in concept, but devilish in the details, time travel for data will happen in the near future, even if just in limited and simplistic ways.
If you've had your review with the big boss, don't let him take new technology evaluation off your goals list. You're going to be called upon to render a verdict on a lot of innovative approaches in the coming year. Some deserve to die, others deserve to be tried and a few scream for quick adoption. And you're just the person to make the call.