IBM Storage indicated it would place greater emphasis on software through its recent Spectrum rebranding initiative...
and a pledge to invest $1 billion over the next five years. The storage software promise matches its 2013 pledge to invest $1 billion on flash storage.
SearchStorage spoke with Jamie Thomas, general manager for storage and software-defined systems in the IBM Systems business unit, to discuss the company's storage software and flash efforts. Here are excerpts of the interview.
Two years ago, IBM announced a $1 billion flash investment and, more recently, a $1 billion software investment in connection with the Spectrum Storage software rebranding. Where will the money go?
Jamie Thomas: Several years ago, we acquired Texas Memory Systems, and that $1 billion [flash investment] was rooted in that acquisition -- acquiring that capability, bringing that in house and creating a flash business around it, which has been very successful for us.
The $1 billion investment in Spectrum Storage is the aggregation of all the capability that we're building and, of course, the applied research that we invest in IBM to support that. The Spectrum Storage family includes both the ability to manage and optimize these storage environments.
Can you expand on what's really going to be new and different?
Thomas: The two newest offerings here are [Spectrum] Scale and [Spectrum] Accelerate. Scale is based on our file system that we invested in over a long period of time, GPFS. But we have invested heavily to add new protocol support to that offering to be able to support objects through OpenStack Swift integration to allow us to support both file and objects. To be able to integrate big database environments effectively with flash and with tape through our LTFS -- what we call Archive Now -- our archive capability for interfacing that with tape is very, very important for the environments where clients just want to be able to look at tape-based data as a big pool of data without moving it around effectively. Lots of new investment has gone into that.
In the Accelerate area, while the XIV is a proven appliance, the key thing here is the innovation required to abstract it from the hardware so that it can run on commodity-based storage that a client may have. That was an investment that we started over a year ago.
The other thing that we are doing with these infrastructure storage products like Scale and Accelerate is putting them on SoftLayer. Last year, we put a pay-as-you-go pricing model on SoftLayer. We have clients out there that are running that asset, particularly for a Hadoop-based data process. If you think about clients that want to do big data crunching, but they don't have all the capacity in-house, this gives them a lot of flexibility. In other words, you could run it on-premises, and you could just scale out automatically into SoftLayer for additional capacity.
One advantage of software-defined storage is being able to run on commodity hardware. Is that a direction you see IBM pushing farther into, where there is a de-emphasis on hardware and greater emphasis on software?
Thomas: That's definitely part of the strategy. That is exactly what assets like Scale and Accelerate do. You can run on commodity-based hardware that you already have, and you can also choose to buy that software asset packaged as an appliance. We give clients an option. Scale itself runs on Intel servers of any kind. It runs on power-based servers as well.
One of our new clients -- China State Grid, the largest energy company in the world -- saw that value in Accelerate specifically. They already had a bunch of Intel servers. Some of these clients have thousands of Intel servers, and they can just take the asset, install on what they have, and it gives them freedom over managing that hardware. If they decide they want to run Accelerate on 25 boxes one day, then fine. But they could use those boxes for something else tomorrow. The box is not tied to that particular software.
Different clients have different needs. Think of managing storage infrastructure software like managing database middleware or application server middleware or any big piece of software. When you manage a database, you the client have to have high availability characteristics around your hardware environment. You have to have 24-by-7. You have to have failover. You have to control that. For those clients that feel like they are good at that kind of thing, because they already run very large IT shops -- there's many, many of those, whether they're service providers, large financial concerns, energy companies -- they then say, 'Well, I just need the software. I can manage the high availability characteristics of the Intel boxes.' There are other clients that say, 'Well, I really need help doing that, so I really want to buy everything as an appliance because in the construct of an appliance, IBM is going to control some of these other aspects.' Our strategy is about allowing the clients to have choice.
Did you sense IBM was losing customers because it didn't offer software-only options that could run on commodity hardware? Are there fears you may lose sales on the hardware side now?
Thomas: I believe the primary motivation was talking to clients who were moving into the next generation of infrastructure, which is clients who are looking at what was really required for big data and analytics and what was really required for private cloud implementations. When you go and talk to those clients and those buyers, you find that the infrastructure was already changing quite a bit. It was an infrastructure that needed a lot more flexibility. It needed the ability to more easily scale out, scale back. And it was amazing to me how much of it was already in play. But if you didn't talk to the right person within an organization, you might not have perceived it.
Now like any business, you have to say, 'Well, if I go in this direction, am I going to hurt my existing business?' That's always a consideration. But, also, you can't deny the future. It's the typical innovator's dilemma. But, we have seen evidence that there's a great combination of these software-led environments with the need to also improve different performance aspects of applications with things like flash. We think that certainly software-defined and flash go quite well together. And if you look at our flash-based offerings, the value of an offering like FlashSystem V840 is the combination of flash with the software.
Do you foresee products like your high-end DS8000 having a higher percentage of flash over time? Or, will IBM start to promote FlashSystem over other systems?
Thomas: No, we're going to do both because frankly our DS8000 with our all-flash enclosure is really designed for mainframe-attached workload environments. The way we've done the interconnectivity between that storage box and the mainframe allows really high-speed connectivity. We've optimized that system for a database workload running on the mainframe. And I had some clients that deployed this in the last part of last year that got extraordinary performance benefits out of that combination. We've invested in all of these new characteristics to make it work well with that mainframe environment, so we're not going to change that.
We believe more of that high end will go to flash over time as they better understand it. There's a natural refresh that occurs with a lot of those products. And as they refresh, I think they'll look more at flash.
What is IBM's strategy around object storage? Does it primarily revolve around the integration with OpenStack Swift?
Thomas: We certainly are invested in taking advantage of the ability to integrate with OpenStack Swift. We've already done that within our Scale offering to allow clients to support objects in the context of that capability. Our cloud environment, the SoftLayer team, already supports object storage through Swift. And we have a lot of organic work that we've always had under way in our research and software delivery teams. There's not anything that we're ready to announce. But certainly, we watch the enterprise object market.
Can you envision IBM having its own offering in the object storage arena?
Thomas: Oh yeah, I certainly can envision it. I'm not prepared to make exact statements about what we may or may not do.
IBM unveils new Spectrum Storage branding
File, block options added to IBM's SoftLayer cloud