 Okay, we're back. This is Dave Vellante. I'm with Wikibon.org. I'm here with John MacArthur and I'm here with Claude Barrera, a cube alum. Welcome back. Hi. Good to see you. It's good to see you again. Claude is a distinguished engineer with IBM, a longtime visionary in the storage community, and a spokesperson for IBM. So good to see you again. IBM Edge, an amazing event that you guys have put together. So congratulations on that. You must be feeling pretty good about it. Well, yeah, it's good to see this many people gathered together to talk about storage and, you know, storage is kind of cool, interesting stuff. So we like talking to customers and we like seeing each other, which is something we get to do in events like this. Yeah, so all storage has always been interesting to us, right? I mean, it's sort of ebbed and flowed over the years, but IBM really reasserted its commitment to storage, I don't know, whatever, five, six, seven years ago. And we're starting to see the fruits of that labor. Talk about that a little bit, that sort of reassertion, that transformation. Well, so one of the things that I think has changed quite a bit in the last maybe three or four years. Ten years ago when we talked about sands and we first got started in storage networking, storage people, at least in IBM and even across the industry, were talking to each other. So we were talking about how to build better RAID stacks and better disk arrays and all that, and it was all really very self-contained. And in a company like IBM where we talk about end-to-end user solutions and software and server stacks, et cetera, we were sort of a bit of an outlier. And what's been happening in the last several years is storage gets more and more integrated into the rest of the system stack. So the question now is how does the whole solution behave? And when you ask the question how does the whole solution behave, storage is along for the ride along with networking and the software stack and security and administration and management, and you add things like the cloud idea, and now we're all sort of in one big pile and we have to work together and cooperate. So storage is much less of an insular discipline in IBM now and probably around the industry than it's ever been. You've been waiting for this day for a while. This is a good development because we present ourselves together in front of the customer and say let's talk about your overall business problem, the economics of the complete solution, not the individual components of it, and what your requirements are. When you think about things like migration of some of the data or all of the data or some of the applications or all of the applications of the cloud, how do you think about what does IBM have to contribute in the area of service level management for those kinds of applications that are going there? Well, so we come at it in probably three or four different ways. That's usually us, right? We're broad and wide. The answer depends, is it depends. So some thoughts. There are cloud instances in which you're trying to solve a very specific problem. So I'm not trying to do all of IT. I'm just trying to do archive well, where I'm trying to provide a backup service to somebody that doesn't want to do their own backup or offsite recovery for somebody that doesn't have a second site. So there are focused solutions around those. And we even work very directly with business partners. There's an outfit in Europe that has been building backup cloud solutions using TSM. So they've basically built their own cloud structure. The TSM is a client that you can download and install if you're the subscriber to this cloud service. And the business partner stands up a TSM server as the cloud service. And then for other environments, it's more comprehensive. I want a full IT stack to run not all IT applications, but a few. I want to be able to host virtual machines and do dev tasks. Or I want to host a few virtual machines and do collaboration. And there's a solution you put together around that. And that outsourcing business is pretty important to IBM. It is and it has been. And one of the big drivers of cloud deployment within IBM is the fact that SO accounts, which had been, we'll take over your data center and run it as a data center because we're better at running data centers than you are, is now morphing into we've got these collections of data centers. Wouldn't it be great if we could cloudify them all, bring together multiple SO accounts and achieve the economies of scale and the variability of asset deployment according to which SO account is demanding more or less. Now there's new requirements to go with that around multi-tenant security, but that's part of the work that has to be done around clouds. And you're also active in business process outsourcing, so going even another step further. Does the information that you sort of, you're able to capture there, drive down into storage architectures and inform the product managers? Can you build those connections or is that too far a reach? Well, no, I would say it's a very important part of how business is done in storage. If you go back all the way, like to the late 60s, that dialogue went on to create database. Basically people at Caterpillar said, gee, we have this need to deal with, they didn't have a good word for it, but we eventually called it transactions. And people on the IBM research side said, oh, we understand more or less what you need. You need to be able to atomically do an operation and guarantee that it works and guarantee that it completely works. So the dialogue now is around things like, how do you want to do analytics? So what transactional data are you receiving real time? And what insights do you need to be able to create real time as opposed to next month or tonight or at some later date? So a transaction comes in, do I honor that transaction now or not? And what role does the infrastructure have to play in supporting that business requirement? So the only way you learn about important business requirements are about talking to the customers that actually have them. So in that vein, I mean, the more we talk about simplifying IT, the more complex IT gets. So what do you see as the big challenges that customers are facing and how are you responding to them? Well, so you're right, we want IT to appear to be simple, and the goal of vendors is actually to create hidden complexity or hidden functionality. So the big problems that I think customers face, in particular vis-à-vis storage, is the fact that they have way more data than they ever thought they were going to have, and they have obligations as to how that data needs to be handled, how long it needs to be kept for regulatory purposes, what security needs to be applied to it, and they have to be able to do all of this affordably. So within the bounds of their economic model, how do I make sure that as data comes in it's not exposed incorrectly, it's kept for the right period of time and deleted after that, and that they obtain whatever business value can be obtained from it in a reasonable way. So it's a complicated equation of there's requirements I must meet, or I shouldn't take on this responsibility of stewardship for data. But there's also business opportunities to be had for having it and using it for improving operational logistics. So probably the single biggest point of discussion nowadays today is around the economics of owning data. The economics of owning data, not managing storage, but owning data. The economics of owning data. Don't give me that data because I don't want the cost of managing it, or give me that data. That's the risk side. It's a reward side. I have data I can improve my business with it. Yeah, so we talk a lot about information asset and liability management, and it seems like the last ten years we've been talking about the liabilities of data. It's big, it's expensive, it's growing, I can't manage it. Compliance, security, privacy, and then all boom, big data comes on the scene. And there's certainly a vision being put forth that the scale is going to tilt toward the asset side. Do you buy that? Where are we in the game of value creation around data becoming more of an asset than a liability? Well, I think we're well down the road of data being an asset. I think there's lots of industries where it's already well in play. Certainly financial services and anything involving logistics and retail. I should clarify. I'm really talking about a step function in that asset value creation. So yes, I totally agree. There's many, many industries that are very data driven. You know, Nicholas Negrapanti, you know, model. But the industry is promising a huge uplift, a huge step function in that value creation. And I'm curious as to... Yeah, data warehousing didn't deliver. Some of these have been around for a while. We've seen a lot of, as John's saying, sort of failed initiatives. Is big data different? Well, so of course the answer is going to be partly we'll see. But I think there's lots of reasons to believe that for a number of problems that society faces, we're going to have to have all the efficiency in our delivery models for food, for clean water, for regulatory compliance across lots of different industries. In our society, it becomes increasingly complex. The idea that people are just going to get this right through smart people looking at printed reports that are three, four, five days old by the time they get looked at just isn't the model any longer. The speed at which modern processes work and the scale of them and the dependency that humans have on these processes working properly requires that we do a better job. And I personally am very confident that the enhancements that come along with data-driven business judgment are going to prove important. Now, one of the leading areas of application of analytics is around things like homeland security. Knowing that there's enough information out there if I were to put the pieces together properly to maybe steal a metaphor from Jeff Jonas. If I were to put the material together properly and derive the insights that I need, I could prevent the next bad thing from happening or fraud from happening. It doesn't have to be disaster scale, it could just be theft. So those kinds of insights derived from data through analytics I think will be commonplace in our future. Yeah, at the same time, when you think about the, we use the homeland security example of how many people get stopped, what's the impact on sort of efficiency of air travel, for example, dealing with the false positives or the false maybes, as opposed to dealing with the positives or the false negatives. So that balance between being able to analyze data quickly and ingest new data, but maintaining privacy such that we can deal with some of those issues. I think that's where some of the big challenges are. Yeah, I agree. And then at a more pedestrian level, there's the pure question of what does all this cost to keep as I acquire data at an enormous rate with very sensitive sensors that collect really large scale objects? What does it cost me to capture those things and to have them sitting around for long periods of time? Yeah, you guys have some announcements that you're doing around real-time data compression. You want to talk a little bit about that? Yeah, so compression's been around for a long time in data storage, but we've really only used it for very cold data. For example, we've been doing tape compression for around 15 years or something like that because the computational overhead associated with doing the compression was viewed to be onerous. You even had a side processor on the mainframe to handle compression. Yeah, hardware compression. So what we're doing now through innovative analytic technology in software is doing real-time compression of online data as it is received by a disk array. So applications in middleware and operating systems are completely unaware that this is happening, but the capacity of the storage is uplifted by a factor of 2 to 2.5 to maybe up to 4, depending on what the data is, with not only no performance degradation, but in some cases even performance improvement, because you're actually writing less data on the slow physical disks. And the rehydration of data from a compressed state, what's that? Actually, the arithmetic compression algorithms, Lempel-Zev in this case, rehydrate very easily, so that part actually is not computationally taxing at all. We were talking earlier a little bit about Flash, and I know in this session before Mark Peters from ESG was talking about Flash a bit. There was a speech that Steve Duplassie gave at EMC World saying, don't worry about quality of Flash, don't worry about write my algorithms better than your algorithm, don't worry about this, all that stuff is going to be taken care of just by Flash, put it everywhere. What's your sort of thinking around Flash? Well, so I know Steve and think he's a lot of fun. When he's saying, don't worry about any of that stuff, I assume his audience is the actual end users. He means Claude's going to worry about it. Yeah, that's right. So what Steve is saying is that the vendors will have spent many hours in the lab making sure this is really all okay. There actually is a lot of work to do, so Flash as a technology was not picked by us enterprise guys. If we, the enterprise community had picked a technology, it would not have been Flash. Flash is, it wears out if you do too many writes to it. The read-write ratio or the read-write speed is imbalanced. You have to reclaim space. You have to reclaim space and do garbage collect. Like the iceberg from storage tech. Thank God you enterprise guys weren't picking it though. So garbage collection is a good technology. That's right. So what Steve is saying is a lot of really smart people will have taken care of that problem. But the fact remains that there is big important work to do to take this componentry that was selected by the tablet guys and the smart phone guys because their requirements are no moving parts and low battery usage and take this technology that really isn't very good at the enterprise job and clothe it with enough other technology. Over provisioning, write elimination, garbage collect and therefore make it a reliable technology at the enterprise level. So when you hear a lot of true blue, no pun intended, but tried and true enterprise guys, so saying that. And then there's the volume and the pricing, right? Do you feel like the economics, the price of flash is going to overwhelm those drawbacks? Or do you expect them? You guys are, I'm sure inventing technologies here a lot about HP Memory Store and others. Will flash be there 10 years from now? It doesn't even matter. Well, it's a little hard to say. There is certainly a window of opportunity for other technologies to supplant flash. And remember what I said about it's the cell phone guys and the tablet guys that do the picking. They don't care about performance and they don't care about write endurance or anything like that because they don't do enough writes. And there's not a lot of performance required in your cell phone. But they do care about cost. And if a new, better technology can come along which costs less, which means the cell guys and the tablet guys will use it and has better write endurance and better read write performance. Now we're talking. So the premise that consumerization will drive this stands. You're not debating that whatsoever. And all problems will be solved in software? The cover up will be done in software. The making it all appear really robust will happen in enterprise software. So a couple other questions. I'm just going to run down them. So to knob or not to knob? What do you think? I'm talking about turning knobs and sort of automation. So there's three schools. There's one school that says zero knobs is the right number. And just put it in. The thing works. I don't mess with it. There is the school that says enough knobs to appropriately pick in the vocabulary of business policy. So I want to be able to turn a knob and say my business policy is my data does not leave my geo or my country because there's laws that say that that must be so. So I don't want you to just say it's all baked in. Don't worry about it. I want some proactive action on my part to say I made this decision. There it is. You know, I can prove that I took act. The third school is placebo knobs where people have steering wheels, but they don't actually do anything. And that's right. That's exactly right. And I think there's plenty of evidence that those things have value. The database people, not to name any names, have been diligently crafting their algorithms for inner diameter versus outer diameter on disk. Even though we've had five layers of storage virtualization between them and the physical disk drives for, you know, 10 years. So they're happy. We're happy. Everybody's happy. That's cool. He thinks he's getting the out of this. That's right. Okay. So where do you stand in that spectrum? I think the, by customer segment, there will be people that say I actually do want some knobs that reflect business policy. And I think there's also the pure appliance model. I think there are people that would say drop it in. It does a known thing. Just get out of the way. But for certain industries, there is a certain vocabulary and a certain set of practices that I think will be long-term endurance. Okay. Another topic, object storage. I mean, quick thoughts there. Object storage, I believe, has found its role in life. So we first started talking about object stores back in, I don't know, 92, 94. There was a little buzz around it, but nobody could really explain what it was for. And I think the thing that we have landed on is it's for Internet scale access with individual level of multi-tenant security for people all over the world to get a data repository. And the confluence of technology seems to be REST APIs, which are nice and simple and have lots of security controls that can be applied to them. And the very simple put-get interface that doesn't involve a lot of risk of exposure or, you know, people might be able to hack into it that are present in some of the NAS protocols. And the use case is now ready. So there are people all over the world that take photos on their cameras or on their digital cameras and want to put it somewhere where it'll be saved if their house floods or burns down. So the use case is there, the technology is there, and the rest of the technology access is also there. Part of that is the metadata of the UNCAPs, which just really opens up a whole new set of value that can be created. Right. How about unified storage? Unified storage, I think, is also a thing whose time has come, and it'll be even yet more unified when people start adding object storage protocols to what they already have with block storage and NAS storage. So that would be your definition of truly unified. Ultra unified. Look, here's your data. You know, you pick whatever vocabulary you like, here it is, tin cans and string, and your box will accept. And my last question is, will all active data be on flash? Of course, I'm going to say, if so, when? So there's certainly one school of thought that says what you need in the long run is active data on something that's fairly high speed, and then the data tub, either low speed, very low cost disk and or tape yet behind that. That's probably a viable model. It remains to be seen what the cost of flash is actually going to turn out to be. Right now, it's still probably 10 to 20X enterprise flash versus spinning disk. If that were to go to 5X, then you're probably ready for a truly two tier model. If you apply enough compression, deduplication and stuff like that, can you? Yeah, and you layer those on top of any media technology you like, and they add value to whatever the media might be. Start to get some economics that might make sense with the added performance. We could do a lot on that. All right, Claude. Really appreciate you coming on. Okay, well, it's always a pleasure to see you guys. Take care. Thank you very much for your perspectives. Claude Barrera, Distinguished Engineer, Chief Technical Strategist for the IBM Storage Portfolio. Great to have you back. Thanks for watching. So keep it right there. We're going to be back to wrap this segment of IBM Edge. We are live. This is theCUBE, SiliconANGLE.TV's continuous coverage from Orlando. Keep it right there.