You must have Adobe Flash Player 7 or above to view this content.See http://www.adobe.com/products/flashplayer to download now.
Download for later:
Listen to this podcast with Tory Skyers and Irwin Teodoro
• Internet Explorer: Right Click > Save Target As
• Firefox: Right Click > Save Link As
[During] our last time together, we were talking about [the definition and benefits of] storage virtualization. One thing that you mentioned really caught my attention, and I thought it important enough that we delve into specifics behind it and maybe do a little role playing again and talk through the details behind management practices from a channel perspective at a prospect and from the consultancy/end user side, what it is that I'm expecting when we talk about management and some of the benefits that storage virtualization brings to the table. [What is] your opinion on what it is around management and how storage virtualization affected your perspective of management?
What we see a lot of times in the channel is a multitude of management tools and the ability to manage data. There was quite a buzz a few years back, and it was more or less ingrained into how storage is managed today through ILM. So you hear a lot [about] ILM and the ability to tier data, classify data so that you can best manage your data. We see more companies moving toward an ILM structure and the ability to say, "This is a high-performing application so therefore I need to put this on a faster-performing disk." Or you have clients that are saying, "This is archive data. In the past, I used to put this on Fibre Channel disk because I used to put everything on one standardization of disk. But now I've classified it, and I can put it on a lower-performing disk, like a SATA drive or something that doesn't have as high performance, that doesn't spin as fast." So when we see a lot of those types of clients or customers that are getting toward an ILM structure or tiering of data, two things come to mind: First, do they have -- or want to get to -- a mature process of how to purchase disk? They have created some sort of categorization of where these things fit, and it helps improve their buying patterns. Secondly, we see that clients aren't aware, they're in an ILM, it's not mature, and they're not quite sure if [an application belongs on] on high-performing disk or whether it should go on a lower-performing disk. So how do they decide? Well, [storage] virtualization helps you with that.
ILM, or information lifecycle management, was a really, really big topic two to two and a half years ago, even up to a year and a half ago. Everybody was talking about it. Everybody was talking about how it was going to change the world and how it was going to reduce your storage expenditure and all of these beautiful things around efficiencies. Today, you don't really hear the term ILM so much anymore, but the topics that you talked about, as far as the management [benefits] behind storage virtualization [goes], it really does seem to be facilitating that entire process of tiering, doesn't it?
Precisely. In the past, you never had the ability to move your data across different storage. There were tools out there that could help you migrate your data, but the ability to [move data from a higher-performing disk] to a lower-performing disk when you don't need it [there] anymore, move it back when you need it -- that never existed. So what we've seen is, for example, at end-of-month processing, there are times that customers do need that high-end processing disk, but then it sits for another month, and then they use it again, the application gets fired up again. So [with] storage virtualization, [you could] maybe not purchase as much Fibre Channel disk, make an investment in some other lower-performing disk and have the ability to move it back and forth as you need to based on the profile of that disk.
From my experience -- we'll pick on IBM and SVC -- essentially what we're doing is aggregating all of our storage requests into one spot. So one thing is going to keep track of all of those requests coming in and out so that way I could better manage and monitor and put metrics behind and report on what's happening in my storage infrastructure, right? Do you see that in the same way?
Absolutely. Storage virtualization allows you, through that single management interface, to capture metrics and certain types of utilization statistics, overall general performance of disk. As long as you're virtualized in front of it with whatever type of disk you're virtualizing, you can have management benchmarks and provide management reporting around disk not even labeled, [for instance], IBM.
What I just heard you say is, "I'm going to save money." So, I'm a consultant and I'm putting together a solution for a client. What I just heard you say is that from my standpoint and my sell-in and my pitch, I'm going to save me money as a consultant so I can turn around and save my client money, and there's some margin to be had in there, isn't there?
Absolutely. Let's preface it by saying that for one to undertake a storage virtualization approach, there is a significant investment upfront. I think the ROI that you get, though, [comes a lot quicker than with] some other investments that you make because of the management [benefits that you gain for] your environment. There are hard costs, of course, that you can achieve, again, if you pattern your disk purchases after some sort of model, [for example], instead of buying all Fibre Channel disk like in the past, you can start creating different patterns around the various forms of performance within disk. That's a hard cost. Secondly, [you gain more in] the soft costs. You could have less of a management [burden] around disk because you don't have to spend time with each of these individual tool sets. Now you have one major interface, one place to manage all your disk.
It's going to take a little more to understand the concepts of [storage] virtualization and what constitutes being moved and what constitutes being virtualized, but at the end [of the] game, it just makes the management so much easier because everything is done through one interface. I think that's one of the major selling points of [storage] virtualization. Not only that, [but] from the channel perspective, you create a more trusted advisor relationship now. In the past, where your business was very transactional because you'd say to your customers, "You need to buy more disk," now you're actually helping them with their over-arching architecture, something that the whole business could leverage. What we've seen in storage virtualization deployments is customers now starting to create buy-back models within their environment. It was almost impossible to track with these separate islands that you had to manage in the past. Now, based on one management interface, [with] some of the features you have with virtualization, you can create buy-back models that make it a lot easier to track and … charge back the business units if they use the disk.
[That's a] perfect segway into what's burning in my mind right now. Let's take a different approach to this particular topic. I'm going to set up an example. So, this is a real-world scenario. I had a client that had a bunch of very, very sensitive, very manhour-intensive data that of course they had backed up and protected but they wanted to migrate it from one platform to another because they found that their requests were being denied more and more by the central purchasing body because they couldn't, dollars and cents-wise, justify the need for the storage. They had a very specific data migration scenario set up. I took a look at some NetApp [systems] in order to facilitate the use of the storage arrays that they had on site, presenting it to their clients in a single-pane-of-glass management scenario. One of the issues that they brought up to me was that they were buying this NetApp for a one-time data migration. It took me almost a month to provide them [with] hard details as to how and why and when data migration happens and why, if you have the ability to do painless or close-to-painless data migration through the use of virtualization that they would be doing it more often, lowering their cost and getting what they need to get done on existing storage vs. having to keep going back to this approval body and getting storage. [What do you see in] your experience from the channel side, on the whole idea of data migration and how many times you'll actually use it in the lifecycle of your storage and equipment?
We're seeing that it is definitely an enabler, and one of the benefits -- I'm not going to say it's one of the main reasons, obviously, but one of the benefits of having a virtualization model. The data mobility options that you have, leveraging array-based disk migration, is probably one of the most unique aspects of [storage] virtualization. For example, we've had clients that [say], "I'm going to invest in this technology, but how many times are we going to move the data?" And the reason why they question that is because it's intrusive. It's something that not a lot of people like doing. With the data mobility, it doesn't matter. You can move it wherever you want. You can move it to HDS disk. If it's on HDS disk and you need to move it to EMC disk, you can do that with [storage] virtualization. A hardware manufacturer had 165 hosts, all connected to EMC disk, from a variety of platforms within EMC, and moved 163 hosts through storage virtualization in three weeks. In the past, that was unheard of because you had to conform to standards from this individual storage platform provider. You had to move connections, add another connection, and it's a very manual process. I'm sure in your practice as well, you've had that same scenario where it's very labor-intensive, a lot of planning is involved, you have to find out where you're going to move it to, where you're going to point it to. It's very cumbersome, very kludgy. But with storage virtualization, it's a matter of the tool set. Provided that you have everything virtualized behind it, you can move it to wherever you want to move it as long as it's supported by the provider.
That's an uphill battle. The statement that you just made is that in your experience and in my experience, we know that this works, right? But it's an uphill battle convincing clients that it works. [With] the client in question that I'm talking about, it took a month to bring them to the VAR's data center, they had to see how this stuff worked, they had to be guaranteed that there was redundancy, because, again, the data they were moving was highly intensive manhour-wise. To produce it would take lifetimes. And they really weren't trying to take any unnecessary risks. That being said, once they saw it work, once they had a proof of concept done onsite, once they actually started using it themselves, all of a sudden, the tactical side of their house turned it strategic. I started getting more calls. The VAR started getting more calls. There were a lot of questions being asked about things that normally wouldn't come from this shop. These guys were really day-to-day tacticians. All of a sudden strategic concepts, strategic ideas, architectures, long-term vision, things like that all of a sudden started coming from these folks [along the lines of] "What can I do now that I don't have to worry about, [for example] where is this LUN filled? Where am I going to put that LUN? How am I going to tier within this structure?" It was very comforting to see that what we deployed worked as advertised.
That's a good point. I have a customer right now, a for-profit educator in the higher-education field, and they had to vacate a data center for a variety of reasons: They were outgrowing the facility, the acquired some other schools [and] were growing by leaps and bounds, [the data center] was too small for where they were at. In the past, they've brought in major consultancies and maybe the vendor in this particular case to help migrate their data center, now, because they have storage virtualization, it is their strategy to move the data center using storage virtualization. Let's not confuse that with consolidation, because that's a different case. They're just moving to a larger facility. But in the past it would have been a two-three-month exercise in how we move the data, there's this dependency and that dependency and if we move this data, this other person who has access to the data doesn't have it anymore. That's not an issue any more.
You probably know better than anyone that that kind of flexibility means dollars. Let's cut to the chase here. That's why we're having this conversation. It's all about the dollars and cents at the end of the day. And, really, what can we do to help our customers generate these dollars and cents and [how can we help] ourselves generate these dollars and cents? … I was one of those people that had to be convinced when I first started down this road to storage virtualization. What convinced me was when I saw the ease at which I can move from one fabric structure to another fabric structure when new technology came to the market. So moving from 2 GB to 4 GB, the flexibility was perfect.
[Flexibility] is important, but not only flexibility [in] your ability to manage the data but flexibility in how you buy [storage]. You don't have to worry about buying disk from the same guy every year or every other year. You can try different things, as long as it's managed by a single pane of glass. … You don't have to buy IBM for the rest of your life, you don't have to standardize on EMC the rest of your life. You can do a little EMC, you can do a little IBM. So it gives options for the customer. They're no longer being steered down this path.
One of the things that I talk about in [a recent] open storage presentation [I did] is vendor lock-in, and you hit the nail on the head. You now have a cost leverage. [With storage virtualization] you can … force better costs into the scenario.
This was first published in September 2009