 Dan Marbs, he's the AVP of Infrastructure Design at Associated Bank. Come on in, you guys are watching The Cube. Thank you so much for joining us. The Cube is SiliconANGLE's flagship broadcast. SiliconANGLE is the worldwide leader in tech event coverage. And we also have Dave Vellante here. I'm here from wikibon.org and it's good to be here in Orlando. We're here with Dan Marbs, who Callie is a big fan. He's been a fan of the show for a long time. Dan said, I got to be on with Callie. You guys switched gears on me over there. We did. So Dan, kind of explain what you do, what Associated Bank is, and just what you guys do over there. Sure. Associated Bank, we are a commercial and personal bank in the Midwest. We have a three-state footprint. We're in Wisconsin and Illinois and Minnesota. We provide full service banking for all of our clients, so loans, mortgages, and the whole financial services package. I am a systems engineer. I work on our design team. I function really as our lead server and storage designer. So we are here sort of giving the good word about Component. We've been a Component customer for a little over five years from taking our journey with them since they were fairly small, and we feel like we've grown up with them in the storage space. So we're just here telling our story. So was it an awkward transition as Component got bought? No, we were excited to see all of the new opportunities with the Dell family that would be open to the Component product line. A little bit scared because we've seen other acquisitions in the past, not necessarily by Dell. Usually the whole acquisition process is a very scary thing for everyone on the other side. You're always worried that the organization doing the purchase of the smart company is going to take them and tear them apart and ruin the fabric of what it was that made the company so successful. With Component, there's really a whole culture surrounding the company. I mean, they're really a small shop, able to be very agile and very mobile because of their small size. That's awesome. And just a great culture of people that they developed. So take us back to, you said five years, you've been working with Component. Take us back five years ago. What was the world like and what were the drivers to move in that direction? We were using primarily DAZ five years ago, and we knew that after one of our recent acquisitions that we needed to provide some more centralized storage primarily for document archiving, all of our old green screen reports and signature cards and all those sorts of documents. So we know we needed some sort of sand solution. We were just scaling out faster than DAZ was really going to allow us to keep going. So we had looked at a couple of vendors. The company that we purchased had some existing sand storage, and that vendor was really more aimed at really massive million-dollar-plus enterprise deployments, and we just weren't really ready for that at that point. So we actually at the end got down to considering the ecological solution and the compelling solution, and we ended up going with the compelling solution. So we started out small. Our original deployment was about 17 terabytes, and over that five years we've grown from one array with that small amount of storage to eight production arrays with about 900 terabytes in total. Wow. Do you have to have people? Did you add people to manage that? At this point we have four engineers who take part in both the sand design and administration activities, but in our time estimates for the last month we figured it's a one-tenth of one FTE total across all systems for all design and administration tasks. There were like 26 hours of time logged in the last month for sand administration, so it really does run itself. I hear these horror stories about people saying they've got spreadsheets of data to manage where this block is, and that's insane. So we just let the system basically run itself, and we learned long ago don't try to outsmart the hardware, just let it do its thing, and somebody needs more space. We click, click, click, and it's done. Your background is not lawn management? No, my background is actually education. I have an undergraduate degree in music education. I was a high school band director for five years before becoming a touring performer, which was the last time I was in Orlando, and now I work in IT. Kelly was like a junior high school band director yesterday at Downtown Disney. Maybe an elementary school band director. Oh yeah, I was doing the chicken dance. I don't know if I was directing rather than I was trying to follow along. Oh, I thought they were mimicking you. No. With junior high schools, that is often the best strategy. Yes. Try not to get trampled. So just to veer off course here a little bit, you said last time you were here in Orlando, are you still performing? I perform a lot now, just not with that show. In the summer of 2002, I was selected to be part of the initial touring cast for an offshoot of the show Blast called Shockwave, and we performed for three months on the American Adventure stage at Epcot, and then did a seven month US tour. Wow, that's a long time to be on the road. After that the show went to London and I ended up going back into education and then eventually transitioned into IT. Wow, what a life you've had. So to go back to the move to Compellent when you guys decided, you said it was really easy obviously, but were there any sticking points? Was there any tough part to that transition? Not to that transition. We've had a few bumps in the road along the way. We figured out at one point, we had tried to take those original arrays that we had and just make them scale up to be very, very, very large. I mean, at one point I think the initial arrays we had, we built them up to about 160 or 170 terabytes each. And the problem is, it's great to have all your eggs in one basket because it's just there. But when the bottom breaks out of the basket, it's problematic. And we fell into a trap where I think a lot of people do only considering I have X amount of gigabytes available and I have X amount of gigabytes of data. This one is less than this one, so I'm good. Without considering the impact of the performance of those different workloads. In 2007 we switched away from tape and we went to all disk to disk backup. And even from a data recovery standpoint, having your backup data on the same spindles as your production data is really stupid. Because if you have some sort of data center failure, you've lost everything. So we've adopted, I think in the computing world in general, we've realized we can't, even with processors, we can't continue to scale up and scale up and scale up. So now we scale out. That's why at this point we have four production arrays in each of our data systems. How do you do your backups? Do you get a snapshot? We actually are using a disk to disk backup third-party product at this point. For some of our very largest systems, like our document archiving system, which is currently around 17 terabytes of live production data, we actually use snapshots and then replicate that off-site to our other... What's your backup software? We're using a product called Evalt right now by i365. You know Evalt and like it? Evalt, I know a little bit, I know i365 a bit. I mean it's sort of an emerging, you know, category of software, right? It's not the classic Symantec Tivoli... Legato base, right? So yeah, I like it. I mean, for disk to disk, right? I mean, that's the future. You know no tape? We only use tapes for our core IBM hardware right now, which was out of scope at the time that we initially implemented that solution. But for all, we're primarily a window shop. For all of our Windows boxes, the backups are all disk... all tape-less. Server virtualization? We are using both. In the data centers, we're primarily using VMware's ESX40 and 401. We are probably 70% virtualized in the data center at this point. We still have a significant physical hardware footprint because we have approximately 300 remote locations, branch offices and brick and mortar back office locations and it's without fully revamping your infrastructure and going with VDIs and thinking about converging all your data inside the data centers, there's really just a need for physical server in these locations. Talk about the bank. What's happening at the bank, at the business? Been talking about different industries. What is that? What's happening? Well, we're really trying to provide service to the customer and give them all the channels that they want. This past weekend, we just launched a major revision to our online banking platform and we now have a mobile banking platform. So as people begin to use their tablets and smartphones as really their primary interface device to the internet, we want them to be able to have their banking activities be able to work on those devices. So you can get your balance as you can transfer funds and do it all from your phone. Yeah, that's been a huge thing on the consumer to use the phone for banking and it's been interesting to watch as those changes happen. So do you see that going anywhere different than it is now? I think as that platform as a smartphone and tablet platform becomes more capable and has more ability, I know there are banks right now experimenting with take a picture of a check and submit that through your mobile banking platform and actually deposit that to your account. We won't mention them, sorry. The bank is not being named. Right. We will continue to see what our partners, our third party providers are able to leverage as those platforms develop. How about this idea of an app store for the enterprise? Is that something that your industry you think we'll get to or your organization? At the point where we're I think fully ready to embrace tablets and smartphones and we really understand the different work types that occur because you think about a loan origination officer versus a teller versus a traveling salesperson versus a traveling IT person and every one of those people interfaces with the network and with the data differently. What we really need to do is continue to be cognizant of what those roles are and adjust their interface to fit. If that's a tablet or a smartphone with an app store, great. If that's a laptop with a Citrix application portal, great. If it's a fat desktop, great. We want to let the business drive the technology decision so we can provide the right solutions to those folks rather than simply trying to shoehorn technology into that we think is cool into a place where it might not be needed. We were talking off camera, you're doing some desktop virtualization, but it hasn't really hit the whole low-mode space. It's a pilot project at this point. We're looking at if you have a large back office location that becomes unusable Tornado, Fire, Flood, Godzilla, what have you. We need some way to allow those users to be functional in as short a time as possible. We've taken some unused space in some of our current locations and put a number of thin clients out there and we're leveraging VDIs right now for workspace recovery and business resumption for those folks. So it really hasn't hit the mobile side of your business yet? Not yet. Do you expect that VDI desktop virtualization, why do you even call it desktop virtualization anymore? We've got all these devices. We have a buzzword and people like buzzwords. Speaking of buzzwords, you do server virtualization and use some VMware and the VDI is VMware as well? Yes, it's on VMware right now, fronted by Citrix. Okay. How do you do disaster recovery? Very carefully. We actually, for a lot of our critical applications, we leverage some of the asynchronous replication features in the compelling product. Our document, our archiving system, for example, we have, I believe, nine or ten virtual servers in a two-node cluster that backends the database and all the file share data and that entire infrastructure lifts up, replicates over to the other data center and can be presented on other hardware in the opposite data center and we take advantage of some network tricks with stretch VLANs and that sort of thing. So you can bring the same servers up in another building on the same IP address so to the client, nothing has changed. Okay. They access the application, their client that they have on their workstation interfaces with the back-end components in exactly the same way and many times they're not even aware it's moved and we can move that entire application which is close to 20 terabytes of data in under two hours. So you said that's an active-active? Yes. But you said it's asynchronous before? We don't have fiber between the two sites Okay. So how do you deal with RPO? We take snapshots of that application every 15 minutes and replicate those across and they sync relatively quickly. It's just with the mechanics of synchronous replication it would put an undue burden on the time to complete a transaction. So you don't have a zero RPO requirement? It's as close to zero as we can get. With our core transaction processing system on the IBM hardware we use their MIMX replication product which is transaction-based replication and with some of our larger SQL databases we do database mirroring between sites because that gets us a lot closer to zero zero RPO. Okay. Do you see that business requirement shrinking? Is 15 minutes okay for the business? It really depends on the classification of the application. We have some which are really like our core transaction and you can't lose anything ever. Right. That's on the IBM hardware that's using the MIMX replication which is transaction-based replication. That's essentially as close to zero RPO as you can get. As you can get. So you said you're here to share at the forum to share your story to interact with people and other customers I guess, right? How's that going? Have you met anybody really interesting? A lot of chatting with analysts and media folks over the last couple of days I think that this conference is really interesting. I think you have people who are from the equal logic camp that are apprehensive and excited all at the same time about compiling it and you have people from the compiling camp who are excited and apprehensive all at the same time about what does the merger into Dell bring and how do these partnerships affect the product. So our experience has been that the shares in the space that we've worked with generally have the customer's best interest in mind and we believe that now with compiling being part of this larger family that it's going to open even more doors. It might be we have to digress we have dear friends who live in England all we have probably seen in close to 10 years that coming over this year we're going to rent a house down at the beach with our kids. Our kids have all grown up. I'm jealous. It's like the two families getting together, the parents all know each other but the kids, you know, it's like it's going on here. It's a good analogy there. I have no doubt it'll all be good. So what about compiling? Let's talk a little bit more about that because I've talked to a lot of compiling customers and actually you guys are kind of boring stuff doesn't break just set it and forget it I mean what doesn't compiling do well that you wished it did well? I really can't think of anything. What is up with that? I don't know. Having been a customer for so long is that I have really good relationship with a lot of the senior engineers and senior support people so even when it's a feature request I just spent some time talking with one of the directors of technical engineering today and I was like it would be great if you replay manager product which is their VSS snapshot engine if it just did this all your PowerShell integration if you could just add these two little things okay I'll probably have them in less than a month wow I'm like yeah this didn't work exactly okay sure I'll fix it. Do you see that changing? I really don't because I really think I continue to be struck in both meeting and listening to Michael Dell just how he really seems to understand the human aspect of this business we talked about the Dell Salesforce becoming trusted advisors and really building long term relationships and I think that really looking at and embracing the co-pilot support model and just understanding what that culture is all about and we're really good friends with the president and he's really confident that that culture is going to be able to stay in place and it's just going to build. They're adding all sorts of staff to the right people that are going to keep that same culture alive. It feels great. It's been on the Cube a couple of times. Yesterday and today here. This week we had them on a VMWare a couple of times and you hear the same theme compelling customers. We had Heineken on at VMworld and absolutely love it. It's hard to believe almost. It is hard to believe. This is IT. Let me tell you the inside of baseball but I think some of the frustration that generally we as technologists have with our technology providers is that we understand that not everything is going to work 100% right 100% of the time but when you get there's another vendor that I work with it took me 6 months to resolve what I consider to be a fairly basic support slash design issue. With compelling I've never had to wait more than a couple of days for RCAs. They escalate it up to the highest levels of engineering and they will get you the answers you need and you're actually in a jam where something hasn't worked right and it happens to be service affecting they will get all hands on deck and fix it as fast as humanly possible. It was interesting when we started Wikibon my colleague David Floyer and I he did a lot of technical analysis and we looked at all the various supplies and we asked each of them can you give us examples of what we call the hero report so the hero report is I'm curious as to whether or not you use it it shows you what your utilization is allocated versus written all the money you're saving do you see those statistics do you use those statistics do you report on that? We don't really consider those I think we have a rough idea of what those numbers are but we don't track them to the penny the way some other organizations do we'd just rather spend the time continuing to improve our design implementing better solutions we can draw them up if we need to The reason I brought it up is because Compellent was really one of the few companies that said oh yeah we have that first the button got it and essentially they took metadata in from all their customers there wasn't customer data it was metadata about how the system was behaving and they just shared it with us we were able to run a statistical sample on the efficiency of the products traditional arrays but you couldn't get that type of reporting out of the other systems and it struck us wow the reason is because it's so simple fundamentally there were some others too but it was sort of the modern architectures that Compellent 3-par was another one that was really good at that and some of the other stuff wasn't so it sort of underscores that whole Cali and I were talking before this event the IT at our home is better than the IT at our work you were asking earlier about better do you mean easier better experience as an IT person you might ruffle some feathers saying that but the reality is it's pretty easy to run IT at home you get your Gmail, your phones there's some complexity but you were asking Cali is that relatively new Compellent is one of those companies that I think catalyzed that whole shift consumerization of IT driving into other parts of your business I think that if you really look at what it means to embrace the concept of private cloud I think Compellent will be a facilitator for us as we really put the rest of the pieces in place with their storage and virtualization and then all the process shift that has to happen to really embrace private cloud so private cloud is that something you guys think about that's a term you use internally? we don't use the term internally there are a lot of things we do that align with what people traditionally believe private cloud to be what makes your private cloud what makes it cloud? we are trying to as much as possible automate away the minutia of system provisioning and capacity addition we really want the brain resources that we hire people because they're smart and they're good at what they do and we want them to be able to use those traits those assets in the most effective way which is not click provision make the busy work go away and let people think about really what can we do to help drive business initiatives like what systems can we play like what high level integration can we do that's going to make all of this make more sense to our customers as IT our customers are both all of the tellers and the other frontline employees and then obviously the customers integration with your financial systems and the business systems that matter and drive value exactly so there's a lot of customers out there that might be using legacy systems that might be using Das what advice would you give to those individuals that are thinking about moving to a more virtualized storage environment I think going back in time and if I could give myself that same advice I would say understand what you want the system to do and how you perceive that it may grow I mean everyone talks about this massive growth of data and how it's been hundreds of percent every year you just have to really understand what that's going to be and how you're going to use the system we now understand that we have to monitor both raw capacity and then the overall performance characteristics of the data and the systems that we're integrating and that's where we failed early in 2007 and 2008 and got ourselves into a bit of a jam so have a vision have a strategic direction it's a lot easier to put technology into a plan than to make a plan after you've already decided on a random bit of technology let the business drivers determine the technology you're going to use rather than forcing technology to drive business strategy because that leads nowhere good that's a good thought to end on we really appreciate your time thank you so much alright, you have anything else before it leaves I just wanted to make sure I didn't cut you off fantastic background very interesting stuff we'll have this up live and this is live now we'll have it up on demand on siliconangle.tv do you have a twitter id? I do, it's at danmarbs okay, m-a-r-b-e-s correct? alright, well tweet me and we'll try and talk a little bit later thank you so much for watching thank you very much