“The enterprise storage market will recover to about the 2008 levels; we think that we will outgrow it”

Laura Guio, Vice President Storage, STG Growth Markets Unit, IBM spoke to Prashant L. Rao about the market for enterprise storage, interoperability, automated storage tiering, deduplication and cloud computing

Laura Guio

If you look at the market for enterprise storage last year, there was a definite slump. Has there been a change this year?

We did really well last year even though there was a depressed market out there. On an annual basis, my growth markets unit alone showed double digit growth in 2009 in total. When customers were purchasing in 2009, they were stingy with their dollars. Rather than just throwing money at a problem they were stepping back and taking a look at it thinking, ‘I need to solve this problem longer term than I've been doing in the past’. Not only can we provide the point product technologies to solve the immediate gap but we can also show the client other things to add to help reduce cost, reduce data—whatever the issue might be for that particular client—give it to him in bite sized chunks and show them the roadmap to say here's where you can go as you do investments further out into the future.

As you head into 2010, we do see the market turning around but at the same time clients aren't losing this thought process that they got into during the tight times. They are saying, ‘Even though we got a bigger influx of capital dollars, I'm going to be a little stingier with it than I was prior to these recessionary times that we faced.’ So we have a lot of that mindset going forward into 2010 as well.

We are predicting that the market will recover to right about the 2008 levels. 2009 took a dip as far as the market is concerned but as far as our expectations go, we think that we will outgrow it.

Interoperability is a vital aspect of computing in today’s enterprise. Tell us about IBM’s efforts in this area.

We have invested in this for a very long time. IBM was one of the co-founders of SNIA. One of the premises of SNIA was that, with all of the different vendors' products that are out there, there has to be some commonality to this. You can't put the customer in the pain point of trying to figure out ‘How do I work with all this stuff that I've got on the floor?’ Early on, we started working with them on adopting SMI-S technologies, common interface protocols between the hardware and any software that's out there. Our TotalStorage Productivity Center (TPC) product, which is our storage management software, adopted this philosophy many years ago. I was running the development teams at that time and we took the attitude of put yourself in the customer's shoes. You are going to have different vendors' products on the floor and you have to be able to maximize all that stuff on the floor and take a seamless view of ‘How do I manage the storage pool from one vantage point and not with tools for every little piece of it?’ Companies like HP and EMC have not only put in proprietary protocols in there but also these common SMI-S interfaces. We manage our own boxes very well but we manage the other vendors' boxes as well and we manage it as a single storage pool—if that's what you want as a customer. When you look at interoperability and you look at the ability to manage a floor of all different vendors' storage devices, we do that very well.

Automated storage tiering is a popular concept nowadays. How does IBM implement this?

We recently announced Easy Tier within our DS8000 series box which autonomically does that within the device itself. We have that in XIV today and it tiers within the box itself. It is important. If customers want this done automatically, to shift workloads for maximum performance within a box, it will do it. Then we also have the ability—when you apply software products on top of it— for them to tier boxes that they may already have. They have made investments in storage devices in the past that we can tier within using SVC and other things to help them tier out that workload.

We have had customers run benchmarks against XIV and other products in the market that are not running SATA drives and they were astounded. We didn't tell them that there were SATA drives in there. We spoke to a customer the other day who said that when he walked into the lab and saw that, he fell over because the performance was equal to what he was getting with his existing equipment.

If you look at deduplication there are two approaches out there—source- and target-based. Which one do you believe is more useful and relevant or do both have their own roles to play?

It depends on what the customer is trying to achieve. We have solutions for both—at the time of data capture we can do deduplication and in our TSM product we can do it from the backup/archive perspective. If you have transfers going on, you really need to do it at the upfront end if it's a long T1 type situation. Our ProtecTIER product, from an acquisition that we did approximately two years ago, gives us an appliance-based deduplication solution.

Which works better in your opinion, appliance- or software-based deduplication?

It depends on the maturity of the market that you are going into and the customer's willingness to look at alternative ways to do it. Many times they had only looked at it in the archiving part of the business at that time taking the deduplication but with the explosion of data costs right now, with it growing at 60% annually, they have to rethink some of that philosophy depending on how they implement it within an ILM philosophy.

There’s a lot of buzz around cloud computing including cloud storage. What’s your take on this?

Clients are just starting to understand what a cloud is. It's a philosophy that is really no different then what we have talked about for some time now of ‘How do you set up your environment to maximize the utilization of all those components that you have within it?’ A lot of vendors have good point products. We have invested a lot of money, technology and interoperability support that's taken all these pieces and made sure that it's working well together. Maybe it's server maximization that they need at one point of time or shifting workloads around the storage devices within the cloud. It also depends upon the industry that we are talking to. We are finding an uptick in retailers wanting to understand more about implementing a cloud. They have big fluctuations in workloads depending on the cyclical buying that a retail outlet may face. They don't want to buy all the compute power and storage that they need during peak times. What they do want is to be able to utilize it as they can and take what's not being used during the off-peak times and resell it somewhere else.

We are seeing more private clouds early on than public clouds. A lot of that's back to human nature and trust. Once the market starts adapting to the private clouds, we will see more public clouds becoming available.

Several vendors are talking about pre-fabricated blocks of storage, servers, networking that are integrated to work together in the best possible manner. Does IBM believe in this concept?

It comes as a tradeoff; customers can try to balance their existing investments or take that out and put a self-contained unit in there. A lot of it is industry and client dependent. I do see some customers saying 'It simplifies my management of this environment by having it all included in one box. I don't have to worry about connectivity between all those other pieces out there.' I'm not seeing a lot of customers going 'I'm not going to buy from you unless I see this particular area'. The lower-end business finds that piece much more attractive on account of it being self-contained. As you get to the mid-sized and higher-end businesses, you have devices as separate entities outside of a frame.

People have been predicting the death of tape as a medium for quite some time now but it appears to be alive and kicking. Why is that?

When you look at infrastructure cost and how do we reduce that, there's nothing more apparent than the data growth issue that clients are facing. With all the things being levied against them—in banking or healthcare they have to keep records for certain periods of time; probably a minute part of that will ever be accessed but they have the requirement to keep that—they are forced into finding low-cost alternatives for saving this data and there is nothing cheaper than tape. If you keep the same data on spinning disk, you have not only the power and cooling cost but also the floor space issues and all of that. They are using a lot of tape for all types of backup. We have seen an uptick in customers talking about tape again. If you look back ten years ago, you heard that tape was dead. Tape didn't die! You have countries like the Netherlands where customers are forced to reduce their power consumption or they are going to be taxed in the following year. They have no choice but to find a cheaper way to keep all that data. Tape's the way to do it.

Encryption used to be a nice-to-have. Is it essential for today’s enterprise?

You have SOX requirements on the banking industry that mandates some disk drive encryption or data is not considered to be safe and secure. From a security perspective, we also see it in software. Especially when you have backup, you have DR site requirements; you have to have some encryption in the stream that is going from site A to site B. It has to be there at both ends. I don't see clients favoring one or the other. In some cases it's being driven by the industry.