The Return Of The Native.

MARK: Hello, Hal. Guess who's back?

HAL: Hey! It's you--Mark Ferelli--the original "Mark" of "I/O With Mark And Hal."

Where've you been?

MARK: I took a swing at market research and missed.

HAL: Huh?

MARK: I thought becoming a consultant would be a logical next-step in my work with mass storage, but I discovered that my heart was really in journalism.

HAL: What did you learn on your sabbatical?

MARK: I took a really close look at storage I/O and picked up a new appreciation for the "path" and not merely for the "device."

HAL: Do you want me to call you "The Pathfinder" now? What's happening with I/O?

MARK: It's in flux.

HAL: That's news?

MARK: Yes. We're in a transition period. We're moving from what might be called "legacy-bus 110," meaning PCI and PCI-X, toward things like USB and FireWire. Not only that, but also the importance of IP as an I/O pipe is rearing its head.

HAL: Well, connectivity is important, of course, but capacity is the most important thing.

MARK: I'm saying connectivity is now a key growth area.

HAL: I'll agree to this extent: the two elements are both cyclical and like sine-curves that are out of phase with one another, they peak at different times. When capacity is all the rage, connectivity, is on the back burner and, when everybody's talking about connectivity, there's not much happening on the capacity front.

MARK: Yes, the same old rollercoaster. Do you think that we seem to be at a plateau with regard to absolute capacity?

HAL: For a while. But even after the curves change back, I'll always lean toward a "more-is-better" storage philosophy. The big number is the one users buy. Most people can't tell the difference between milliseconds and microseconds in the throughput specs and, anyway, with chip prices so low, cache is king.

MARK: Of course, capacity is a paramount issue, but only if the end-user is adept at managing it. Not everyone has internalized the disciplines of planning capacities, balancing loads, and coordinating scalability. It's not even certain that the future of backup is in tape or online disk. Promoting storage management will take time.

HAL: I won't dispute that, but I do think that having more capacity than you need relieves users of the need to take on such managerial burdens. As long as there's room in a disk or on a tape, you don't have to optimize your storage space and you can buy more capacity any time of day or night over the Internet, now. Not only hardware, but also dial-up storage and archiving services.

MARK: Shhhh! Don't tell anybody or you'll put hundreds of consultants and ISVs on the street! Yet seriously, isn't there an absolute capacity-saturation point for every storage strategy?

HAL: Yes.

MARK: Well, that's why I say that I/O is so important.

HAL: Wait a second. You just got me to agree with you, flat out!

MARK: I learned SOMETHING while I was a consultant.

HAL: Go on.

MARK: Simply adding incremental capacity has an undesirable side effect: can you spell "performance degradation?"

HAL: Awwww. I haven't heard anyone say "Can you spell. in a long time. Would you say, "Everything old is new again?" Just for old-time's sake?

MARK: Let's get back on track. One of the interesting elements that's new in I/O is a change in mindset. We're seeing "storage-thinking" and "network-thinking" coming together. Storage and network hardware are getting so tightly coupled that, pretty soon, you won't be able to separate them.

HAL: I bet that suits the networking companies just fine. They've been looking for a follow-on product line ever since the server-market curve started to flatten out.

MARK: Did you think they were promoting SANs just for fun?

HAL: They're betting the farm on SANs! The days when every network had to be custom-configured are numbered and building "Storage Area Networks" is one of the few ways left for honing one's skills. Networking is getting to be like the disk and tape drive business: you can now buy servers--any size, any configuration--by mail order and ecommerce. It wouldn't surprise me if, in a few years, people will be able to say they're in the storage business and yet never handle any hardware at all!

MARK: Welcome to the 21st century. We'll have to be ready for it. Take a look at that Javabased implementation, Jiro. If it lives up to its promise, all those storage servers and their endless variety of APIs will have common ground to operate on, a separate management layer. The management issue is still significant, even though some hardware is becoming a commodity.

HAL: I know the HDD manufacturers are worried about that trend.

MARK: I think the future of the HDD will be in the new applications that are sweeping the marketplace. In any computer equipment investment--including storage--applications are still king.

HAL: Yes. Users still have to know what to do with that hardware once it's unpacked.

MARK: Yet I'd suggest that, one day, storage management will become just one more app in a network's configuration. The externalization of storage is turning storage subsystems into something of a subnet.

HAL: On that day, you and I will be out of work too.

MARK: You should have a little more confidence in the ability of the industry to innovate. That, for us, represents job security, but I wonder if our readers are concerned about the issue. Let's ask. If you think there's more progress to be made and new techniques coming along for hands-on storage management, e-mail me at [e-mail].

HAL: Yet if you think automation is already breathing down the neck of storage managers, e-mail me at [e-mail].