News Stay informed about the latest enterprise technology news and product updates.

IBM chief engineer Barrera talks EMC-Cisco, XIV and solid-state drives

Clod Barrera, chief technical strategist for IBM storage, discusses IBM's plans for a pre-integrated virtual stack, the future of the high-end disk array product line, and Flash.

With virtualization and the cloud taking a larger role in data storage strategies, vendors are forging new alliances and approaches to the way they deliver technology. IBM has been at the center of several of those competitive and technical battles, and its chief technical strategist Clod Barrera sat down with recently to discuss IBM's plans for offering a vertically integrated virtualization stack to compete with the Acadia alliance between EMC Corp. and Cisco Systems Inc.; management software for solid-state drives (SSDs); and the future of its XIV Storage System and DS8000 high-end disk arrays. What's IBM's reaction to the new Virtual Computing Environment (VCE) coalition between VMware, Cisco and EMC? 

Barrera: It's a recognition that these are areas where there's important growth going on in the world, and that integration of technology in front of the customer is going to be an important thing as people build out these environments everyone's expecting or talking about. You have to do one of two things to be successful -- either you have to be IBM or you have to emulate IBM. So what we see happening is companies starting to try to aggregate together collections of technologies and if they can't do that as a single enterprise, they have to do it as sort of a virtual corporation. The announcements that have been made are codifying activity and behavior we've seen going on for a while now.

At times there's been speculation about does XIV take over for the DS8000 or is the DS8000 dead? The answer is absolutely not. The DS8000 is alive and well.
Clod Barrera
Chief technical strategistIBM There are fears that pre-integrated stacks will curtail customer choice. What are your thoughts on that?

Barrera: There's no question that aggregating things together eliminates choice at the individual component level. I think the path that we've been on around the industry for maybe the last 10 years or so is a very disaggregated view. If you go back to early IBM mainframe systems, they were aggregated things – a collection of technology that solved a business problem. After a while, people became enamored with buying adapters, enclosures, etc., and putting them together. There are people who do that well, like Google, but most people don't.

If you're looking for answers to pressing questions that are driving customer angst about whether their infrastructure's right or if they're on the right path, it's getting harder and harder to imagine it's going to be done by buying a whole bunch of components and assembling them.

The balance, I think, is to acknowledge the existence of the best possible building blocks. Let's not debate whether Ethernet is a good thing; let's use standard building blocks to deliver a business solution for a client. In the course of doing that, vendors will have to make choices about what gets built and aggregated together, and what's allowed to be used and what's disallowed. There's surrender of some degree of choice, but I think if the vendors do this smartly, clients will look at it and say 'They use all the right building blocks, the cost structure is reasonable, and I get a solution to my business problem.' IBM came out with its own pre-integrated products in October. How do you see these integrated data center stacks evolving?

Barrera: IBM is focused on, first of all, the idea of introducing solid-state disks into our technology stack. Not just 'Here's the right drive and a place to plug it in,' but integrating up and down the entire management stack. SSDs are great performers for IOPS, although not all that much better than regular disks for sequential performance. They're expensive and you have to be smart about where data is placed. Management complexity is a bad thing, and adding technology that increases management complexity is really bad. This will be a major focus area for IBM -- tooling that lets you identify the need for better performance for particular applications or volumes.

The announcement you saw for an IBM intelligent archive in [last month's] analyst meeting -- that's just the beginning of a path. I think it is going to be a big important platform for us. Over time we'll be adding additional management capability, additional ingest methods and devices supported within the archive.

A major initiative within IBM and elsewhere around the industry is the notion of integrating virtual storage and virtual servers and network and workload management into a single comprehensive information infrastructure. One of the notions everybody is sort of signing up for is the idea that infrastructures have to be dynamically responsive to workloads and [that] workloads in particular data centers can change dramatically -- shift to shift, or even hour to hour -- so the idea of integration of storage and virtual servers is going to be a major piece of work.

Virtual servers have different characteristics in usage of storage and management than physical servers. Connecting a server used to be done with a guy with a screwdriver and cables, and it stayed provisioned for a long time. Now it may move from hour to hour, and the associations with storage have to be preserved or torn down and stood up every time that happens. In order to get real value from the virtual data center environment picture everyone imagines, which we call dynamic infrastructure, new plumbing is required. So will IBM deliver an Acadia-like stack?

Barrera: I'm not sure that I'd call it Acadia-like, but we're working on the problem I just alluded to. There's been a lot of talk about the positioning of IBM's XIV Storage System and DS8000 series arrays lately. With a new XIV version just released, can you talk about how that positioning has developed so far and where it's headed in the future?

Barrera: There are a number of different segments within the disk market and we've had kind of an open mind about what is the right place and positioning for XIV technology or derivatives that we can create from the XIV box. To date, we have a lot of success selling XIV into accounts interested in aggregating tier 2. We're looking at XIV as companion storage for a number of other workloads we think are growing, like archive, cloud storage and scale-out file systems.

At various times there's been speculation about does XIV take over for the DS8000 or is the DS8000 dead? The answer is absolutely not. The DS8000 is alive and well, and we just made a big incremental step in its performance [with the DS8700]. DS8000 tends to take a performance step then a functionality step; we don't do them together in terms of optimizing our own tests. Next year as we get into the spring you'll see functional improvements to the 8000. In our minds there's no big rush to make the big-box disk environment into a single thing. How has the EMC/Data Domain acquisition impacted the competitive environment for IBM's ProtecTier?

Barrera: Imitation is the sincerest form of flattery. EMC and NetApp both sort of woke up about same time and realized this dedupe stuff, particularly in backup and large archiving environments, is going to matter a lot and went hunting for a company. We went through that exercise a couple years ago and landed on Diligent. What about EMC's earlier acquisition of Avamar?

Barrera: Avamar isn't a big pervasive player in this area. Data Domain is already a known quantity in the industry. Under EMC's wing they will be brought in more regularly and, over time, I would expect to see deeper integration between rest of the EMC product line and Data Domain.


Dig Deeper on Data storage management

Start the conversation

Send me notifications when other members comment.

Please create a username to comment.