What strategies are on IBM's storage agenda this year?
We have the storage systems group and the software group. In the storage systems group, the focus is on hardware products. We formed a software group because we, and our competitors, recognized the value of moving into software.
We're focusing on storage infrastructure software. Tivoli focuses on storage management. We need both. If you think about an automobile in the 1930's, it was this clunkety thing, and we said, we need better management tools. Let's put lots of dials on the dashboard like a tachometer. We could also say, why don't we build an automatic transmission? Let's go change the infrastructure so it's easier. We're in the automatic transmission business. Tivoli is in the management business.
Last year, we were hearing that simplified storage management was the key to improved storage utilization. Now, you're saying the focus should be on improving the storage infrastructure. What brought this change about?
If you went back a year ago, when people said storage software, what they really meant was storage management software. This year, for the first time, analysts like Gartner said there's something else going on here. We really have to separate this three ways. One is traditional storage management software, which they broke up into storage resource management and data management, such as backup, archive and hierarchical storage management. And they said there's also a third category
Storage Tank is one of them and our virtualization engine is the other. Those products will be available in 2003. They're in internal testing and will be moving through those stages very soon.
It's really three pieces: Virtualization, common file systems (SAN-based file systems) and standards-based management. We're building a virtualization engine for blocks, and Storage Tank, which provides a SAN-wide file system. We're doing all this in the context of industry-standards-based management and doing it in such a way that our storage managers, Tivoli or somebody else can manage the automation products we're building. If we've done our job right, we won't know who is managing our products.
What are some of the top concerns you hope to address for storage managers?
It's very simple -- it's administrative productivity. A couple of years ago people said the storage management problem is very low utilization. We don't hear that too much today. What you hear today, I believe, is hey it's all just a mess, it's too complicated, all this multi-vendor mumbo jumbo doesn't interoperate the way it should. All of which translates into low productivity for storage administrators. What benefits will users see in the early product releases?
The big benefit of virtualization is storage administrator productivity. Literally separate the application from physical storage. The second big benefit is that it then gives you a platform for applications. You'll have better disaster recovery systems and better hierarchical storage management.
The first release of the virtualization engine offers the basic virtualization ability, flash copy and remote copy, migration services and ability to import data. From that we'll move into more advanced disaster recovery services and hierarchical storage management.
Storage Tank will allow people to consolidate what's in some cases thousands of file servers. I think of it sometimes as NAS for the enterprise. Network attached storage does that. It lets a whole bunch of servers access the file system, provides heterogeneous file support, file sharing, but it doesn't play very well in the enterprise where you need high performance. It starts out in the early release supporting Sun, Solaris, HP/UX, AIX, Linux and NT so all of those systems can share a common file system complete with N+1 failover. We will add to that 390 connectivity in a future release. It's a way to provide the single file system, common naming system, heterogeneous file sharing, but to do all of that with local file system performance.
Storage Tank will be a very complete solution when we deliver it because it's been in development since 1999.
How will industry standards play a role in the adoption of these technologies?
Two things. If you look at the Internet, it works because of standards that are actually very simple. So if you and I read the standards, there's a very good chance we'll read the same thing, interpret it the same way. That's why it's called a simple network management protocol or a simple mail transfer protocol. That means we can all interpret it and build to it. And we'll see the same thing in storage.
The other good news is that all of these standards for interoperability that have come out of SNIA are actually, I think, very, very well done. And they're all based in XML. So, it's very simple.
With a focus on interoperability, vendors are pretty much looking forward to CIM standards. I think in a year, they'll be here. It amazes me that some vendors still try to argue that it's a few years away -- it just isn't. It's going to happen.
FOR MORE INFORMATION