Andrew Monshaw, general manager of IBM's Storage Systems and Technology Group for the past year, spoke with Computerworld about a variety of storage issues. He said he sees no end in sight for tape storage because of its value proposition, feels that virtualization is king and is still unsure where Microsoft's venture into storage is headed. Monshaw also said the open-source software development group Aperi, formed in October with eight other vendors, should produce the first version of a common storage management platform by mid-2006. That platform is designed to allow users to manage storage systems from multiple vendors through a single interface.
What's changed from when you began this job?
No. 1, I think we're at a time when we're seeing a tremendous tail wind and we're going to see many, many quarters of really strong growth -- well more than what's predicted by analysts right now. There's a tidal wave of data coming at people, and customers are figuring out the more data you can capture and use, the more you can turn into information.
After my first year, our vision on information-on-demand is resonating. The strategy is working and validated by customers. I think one of the reasons it is working is customers want end-to-end solutions. They want a company that can bring the right kinds of services and business consulting on the front, the right kinds of product innovation, complementary integrated software and integration with servers. They don't want a patchwork of acquisitions. The third piece is that virtualization is real this [past] year. We're adding four to five customers a day on our storage virtualization offering [SAN Volume Controller]. It's turned out to be a very different business model than what we originally expected and presented to the corporation.
Five years ago, we thought people would sell virtualization and what's turned out is you sell the benefits of virtualization. It's a subtle difference, but from a business model it's a very big difference. When you're set in an infrastructure, it's a really expensive proposition to move out of because of the complexity [not just] of migrating arrays, but all the way through to managing the new stuff and having to retrain people. Virtualization opens this thing up and gives the customers a lot more choice on how they want to spend their money and with whom. It gives them choice. SVC works with just about everybody's arrays and 50 to 60 operating systems.
You said in April that SAN Volume Controller had broken the 1,000-customer mark. Where is it today?
We have about 1,600 customers. We're adding four to five customers a day.
Are you seeing it being used for one specific reason over another?
Almost every customer uses it for [data] migration and uses it for utilization improvements. Interestingly, a high percentage of customers don't even know how much storage they have in their enterprise. Let's just start there. So a lot of this is geared around getting control of their infrastructure and then utilizing it. The biggest headache right off the bat is any kind of migration for an array. Migrations take forever. It takes a long time to plan and a long time to implement. With virtualization technology, it's a snap. Virtualization shields the application from the physical array. It doesn't care that you've moved a physical piece of hardware in and out.
This [past] year has been the year of validation. NetApp renamed everything "V-Series." EMC came out and said, "We were just kidding, virtualization is important, and we're going to try to get this Invista thing out the door." So it's been validated by our competitors. We have customers running up to 700TB behind [SAN Volume Controller]. It is real. I've got hundreds of CIOs that will reference this thing. Now what do we need to do to really accelerate this thing? We're investing 25 percent more going into 2006 around interoperability and new features and functions. The next wave here will be moving a lot of the replication services into the network.
Are you considering moving SVC's virtualization functionality into the network, say onto a switch?
I think there's going to be a lot of implementations around this. I don't want to go into our unannounced product plans, but I do think there's room for virtualization technologies in the network, in the arrays and in the switches. And there will be different implementations that will solve different problems.
Can you elaborate?
EMC's Invista will do what really large, entrenched EMC customers need it to do. So it will provide some migration capabilities, and maybe some limited real, true virtualization capabilities. I don't see it moving to replication services, because they just have too much profit locked up with the way they charge customers for replication today.
High-end storage vs. low- and midtier systems. What kind of shift are you seeing in purchases?
Our average high-end system is about twice the size per average system than it was a year ago -- about 25TB. So high-end systems serve a certain kind of purpose, and we're seeing a lot of consolidation benefits there. Also, I think we're getting uptake in our high-end systems because of LPar [logical partitions] capabilities. We have logical partitioning inside of the mainframe system, called the DS8000, and we're right on the verge of exploiting that technology in terms of running applications on these LPars. We demoed that you can run DB2 on an LPar. Why would you want to do that? Because there's aspects of data collection engines that consume expensive processors. If you can move it to the LPars and run it next to the storage, and then eventually make the storage connections with the application's higher bandwidth, it's a huge cost benefit to the customer. Midrange is just exploding, both for Tier 2 storage and for small to medium-size enterprises.
Are you finding midrange systems cannibalize high-end systems sales?
I don't see cannibalization. I see demand for both areas. Another way to think about that is could robust, clustered midrange systems be a movement in the future? Very possibly. But there will be requirements for large mainframe systems and requirements for midrange systems. But the more important thing in my point of view is that everything interoperates.
Are disk systems eating into tape system sales?
We've probably had our best tape [sales] year in years. Innovations in tape is what will continue to drive tape going forward. The key is customers have to save a lot of this data. There's a lot of compliance regulations driving this. There are tons and tons of images. Our focus has been to stay on or better the 10:1 [price differential] between [serial ATA disk drives] and tape. As long as we stay on that path, I think we have some breakthroughs that are just jaw-dropping. The density and price curves that we plot keep us at a good differential from disk for as far as we can see.
Lastly, the key to this whole thing is to get end-to-end encryption right. I believe we have the right strategy for that.
And what's that strategy?
Link it to the zSeries. Because the hard part is not encryption of the tape. The hard part is the key management. It's how you allow access to the stuff. How do you define access. The zSeries is the best in the industry at doing that.
When do you expect to have that end-to-end capability?
Pieces will roll out over the next two years.
Are you considering reselling Decru's DataFort encryption appliance technology through your OEM partnership with Network Appliance?
We are evaluating this all the time. Our partnership with NetApp is off to a great start. We're having ongoing discussions about how to leverage each other's technology all the time.
You've tried to go to market with iSCSI in several iterations. What's your road map?
We have it on our DS300 xSeries attached box. We also have iSCSI capabilities through our NetApp partnership. We're continuing to evaluate whether or not it has enterprise usability. Right now, the iSCSI market is pretty small and it's stayed pretty small. It continues to be predicted to grow really fast and yet it is still pretty small. However, it's pretty logical that it could be a viable alternative for customers in the future. It's a topic of discussion at every road map meeting we have in terms of our development. However, right now our course is the DS300 and its follow-ons and the NetApp [reseller agreement] primarily.
How has Microsoft's expansion into storage affected your plans?
We're keeping an eye on this space. It's clearly a vendor we need to continue to partner with. Their road maps are not really clear to us yet. We had launched a product in the xSeries space, but we didn't see tremendous uptake there. It's clear they're more active in the storage community, but it's not really clear where they're going yet. We're certainly not dismissing them. I have a lot of competitors in this space and they're certainly on the radar screen. When Microsoft puts their mind to something, they get it done. We've seen more co-opetition this past year than I think we've seen in years. You just have to keep your eyes on who's your competitor and who's your partners, and sometimes they're the same company.
Where are you at with the development of an open-source storage management platform through Aperi?
Have you added any new partners to the original eight? No. But I'm meeting with a potential partner [Thursday]. There's a lot of interest. There is another group, not the top five, but the next five, that are on the sidelines taking a wait-and-see approach. A bunch of us get it and a few people don't get it. They're not looking at it from the customer's point of view. They're looking at it from their point of view. I think enough pressure in the system will force them to look at it from the customer's point of view.
Does this preclude you from participating in SMI-S?
No. But, this builds one common base that does the basic things a customer needs to get started, like discovery and what are the resources in your infrastructure. Then if you decide on EMC, that's cool. EMC should put their investment not on the base, but how do you optimize the management of EMC things. Or how do you optimize the management of Cisco switches, or Brocade or HP or [Hitachi Data Systems]. I think there's some good code that will be donated to this.
When do you expect to see some product from this?
By the middle of 2006 we should be making some very good progress. The board will define the code that's been donated, and then we'll be off to the races. The committee's met a bunch of times. They're defining the governance model and they're making really good progress.
Do you see acquisitions playing heavily into your product road map in 2006?
We did the NetApp agreement and the Aperi thing. Now it's time to get into the OEM alliance and acquisition space. We've built a new team and we're engaged with over 30 potential spaces where this would make sense. So I'd say in 2006 this will be one of the space that will be pretty hot for us.
Would NetApp be a potential purchase for you?
I'm not going to comment on that.