Predicting the Future with Dave Hitz of NetApp

the title of my talk is predicting the future which I'm not actually going to do but I'll talk about it before I do that though I just wanted to acknowledge something that up has been in the news lately last week the headlines were something about layoffs which is not the way that you normally want to be in the news and the word transformation and someone sent me this cartoon and it was a picture of people at a white board and it said restructuring layoffs Trenton chanted like all these different words and then finally at the bottom it was transformation we haven't tried that one yet and that caption was I've been transformed and so what does this stuff mean what's going on we've talked about doing a transformation of our business and as part of that we did do some layoffs last week but I'd like to give a little bit more perspective about what's going on because I think it does help you understand how we're thinking about the future if you look at Annette app historically we have done a lot of kinds of innovating but we have not done very much of a focus on our business processes and just how we get things done and how we operate in the market and that is really something that our CEO is very focused on right now we got a new CEO about 10 months ago and he came out of Cisco before that he'd been at McKinsey there's George curry and obviously and he's got an interesting perspective on what's going on in the market and how do you think about it and I think that sets the stage a little bit for some of what we're doing in that Apple and in fact sets the stage even for the acquisition of SolidFire so his view is if you look at IT roughly speaking it's flattish market overall but with pockets of opportunity and the main thing that you would why you would know it's flattish I mean if you just think back in the 90s every CIO everywhere was asking this question how could I spend a little more on IT to grow faster right I mean that was the sort of mental model if only I invested more here it would be good there now every CIO everywhere is saying basically I wonder if there's a way I could spend less and still get the same amount done or possibly more so so just based on that I see as a whole is going to be flattish but with really interesting pockets of opportunity for growth like all flash arrays and cloud and I'm in a variety of a very so if you look at that George's mental model of the company is let's look at the company as a whole and say how do we structure this company as a whole to be healthy in a flattish environment either because the economy's down or the bankers do something or or whatever but at the same time look for those pockets of growth and figure out how to take advantage of them so that if you do they're gravy that you're not relying on that as a business model so that's that's what's going on there and we've had people say well gee if you think that it's flat isn't why are you doing something like acquiring solid fire and the answer is because that's one of the really interesting hot zones so with that I would like to I would like to talk about predicting the future one of my favorite quotes on predicting the future came from William Gibson the science fiction writer they've given this group I'm assuming you mostly so the line he used is the future is already here it's just not very evenly distributed the future is already here it's just not very evenly distributed and my interpretation of what he meant by that was the stuff that everybody's going to be doing five or ten years out some crazy persons already doing it right and in fact I would argue that a lot of your goal is to go out and talk to those crazy people and figure out well what is that but that doesn't that's not the same as saying go find any crazy person and copy them right there's this whole thing of like which are the ones that make sense in which aren't my personal view is that spotting hot trends is one of the most interesting ways of sort of seeing where the future's going and like I said before the the two biggest trends that I see right now are are especially with respect to storage our flash and and cloud cloud hybrid cloud that whole zone so that's where the bulk of my time I'll be talking about but before I do I'd like to talk about this concept I have of a tenure trend because when I say trend I mean a particular type of trend there's a lot of trends you know you see CIO magazine and there's a list of the top 10 trends for the year and then the next year the magazine comes out and there's a different list of top 10 trends for me those are not that interesting I mean if it's only on the list for a year that's not so much a trend as an event it's the ones that are on for a long time and just to give you an example of the kind of thing that I'm that I'm thinking of IBM introduced the PC in 1982 and it was on the order of 10-15 years before you could just assume that everybody's desk had a PC on it you know it like it was a long you don't you don't have to be a genius and call those trends ahead you don't even need to call them super quick you just need to notice gee we're two or three years into this thing and it looks like it's going to take a long long time another example is it's things getting networked when those pcs first all got installed on everybody's desk they weren't wired together sneakernet anybody and and you know cisco started in the early mid 90s but it really a lot of the work of the I sorry the early mid 80s I mean but a lot of the work of the 1990s was wiring all of those pcs together so so I do think that you can look at trends in these long ways so when I look at trends that have this sort of long-term potential VMware was another one I mean mid 2000s VMware got introduced but it has been you know a long period where more and more and more stuff got sucked into the VMware model so so two of the trends that I see flash and cloud and both of them are really important and one of the things Joe will talk about joke Arizona in a bit is data fabric and how we're investing in flash and cloud one definition of data fabric to just get started would be have got a bunch of data on disk now there's this hot new stuff called flash and plus maybe I'd rather be in the cloud how do I keep my arms around my data whichever of those places it should appropriately be and we could get all fancy about protocol names and definitions and how you're tracking it but if you just think of it as like where the data lives is a complicated hard question that changes over time the data fabric is the thing that helps you get the data to where you want it and maybe back that and that's the kind of the big picture of data fabric that's a pretty good definition so so flash cloud I'm going to start with flash and then I'll dig in on cloud so the the summary of NetApp's position in flash is we actually started very early with caching but not with all flash arrays and we got off to a slow start with all flash arrays so as a result of that we let other folks like like pure get into the market and but that was then I would say right now our flash portfolio is very very strong especially post SolidFire acquisition but even before the e-series in the all flash has let me show you the numbers from our last quarter not including the SolidFire numbers for our all flash business our all flash business was a run rate of 600 million dollars and a quarter over quarter growth of sixty percent that's a pretty big business not quite a billion 60 percent quarter-over-quarter growth rate and so I have to say one of the questions that I've gotten post SolidFire is so you know I SolidFire does that mean you guys don't like what you were already doing are you gonna kill the all flash fast oh man that I will say that just drives me bonkers have any of you guys heard of a company killing something that was six hundred million dollar run rate growing sixty percent from the previous quarter I mean I if can I ask you a question yeah initially when you put your Flash into the products it was done as a caching as a caching process and I'd and I seem to remember there was a discussion that that was a better tech that was a bit of strategy about it philosophy philosophical way of going forward then than having cache sorry been having flash as a tier of storage what was the point that that was changed and that philosophy sort of moved on to to a different level of things yeah let me look a night or was there a point it so here's the let me make an observation I have an eight-year-old daughter when she was four I told her do not cross the street yourself and then later I told her you know what now I want you to cross the street yourself and the question was I wrong then or was am I wrong now and the answer is the situation change I would say in the early day there's two ways to look at flash you can either look at it as slow DRAM but boy is it cheap or you can say it's fast disk but boy is it expensive right I mean that's essentially are you using it as cash or are using it as the storage the off flash array at the time that we started EMC went the all flash approach we went the caching approach their box was so expensive as essentially nobody bought it we were getting small amounts 1% 5% of total storage in flash we actually could drive the cost of the solution down and let people use SATA drives and and get higher total performance lower total cost at that era given the flash prices then I still believe that was the right strategy the mistake we made was as flash prices came down and down and down and people started saying well what about all flash we were like bathing in our own bathwater on hey we convinced everyone caching is better because it was but it became not better and and that's when I said we were slow off the mark on all flash I think it's ironic I believe it was actually our success in flash as a we sold petabytes of flash before anybody else I believe it was our success that blinded us to the change that was going on does that make sense absolutely does that sound like I'm just explaining what it sounds like you've been very honest so have we seen the tipping point now are you guys now selling more flash run spinning discs when so I think there's always gonna be two ways to answer that is it more dollars and is it more capacity oh it's not more dollars yet but it's sure moving fast like I said are all flash business was about six hundred million dollar run rate and if you look we're you know a six five six billion dollar company so no it's not and I mean there's other stuff but but it's it's not it's not mostly the I forget when the crossover point was for tape you know tape was gonna die die die and but it kept not dying right and especially if you looked by capacity capacity still enormous but all the data anybody cares about moved from tape to disc that's what I think is happening with flash the question of is it important to bytes of disk or more gigabytes of flash total industry shipped I think is a less interesting question than saying what's going on in different parts of the data center and when I think when you talk to people about the parts of the data center they care about the parts that they're running the business on the parts that their performance managing tuning doing anything with that stuff is going flash very quickly I will say when I talk to customer panels so you guys here I'm guessing all complete flash converts bought your last disk drive everything should be flash is that a my kind of not really no okay no paint a couple of you yeah I still think there's definitely a place for disk especially talking about cheap and deep capacity where performance isn't as critical hybrid storage arrays are still the sweet spot in the market in a big way I mean if you look at metrics for people who've actually adopted it's just a sliver that have got all flash doll I mean this is still King in the data center well so that was really the point I was gonna make when I talked to customer sessions and I go out and do we call them executive briefings on the road instead of making them flat in that app you know someone like me goes out and we get a bunch of them I've been in LA recently I was in a Colorado not too long ago and we did one in Vegas when I do a show of hands on how many people have a mission-critical app in an all flash array and just do that around I have seen as high as half but that's really rare usually usually it's not very many people have any yet you know 10 or 30% you know so the reason I'm raising this is because it's complicated for me to get up and talk about look all of the data needs to move to flash and our vision as everything's going flash when everyone in the audience is not even raising their hand that they've started and that links back to why I talk about this tenures trend thing because if you believe hey flash is gonna be important and it's this teeny percent and you look at us having three flash platforms in our portfolio the E Series and the all flash faz and SolidFire and you go what are you guys crazy like why well we believe that the world's going to a hundred percent flash and the center of the data center where performance matters and and people just go well that does not make any sense if you talk about it as a ten year trend this is the trajectory where people are headed on no it's not immediate but the curves keep coming down that the deduplication compression ratios keep getting better the stuff you can do with flash and not disk right it you put it in that ten-year context and then you can have I think a more clear conversation so even even with what I just said about all flash fast mean well we we've got three products in our portfolio and the question is that too many how do they fit together and let me describe what I see is the place for each one so the e-series anybody's seen on the web those those racecars made out of welded aluminum tubes the atom or something like that yeah you've seen it it's this hot little car I mean but if you want a windshield buy a helmet with a visor if you want a door there's no door I mean you just like hoister that is the e-series this thing does not have any data services it's not mean it's got protection it's got the basic raid but but a higher level it's not doing compression it's not doing D dupe it's it's a very basic simple and it goes like a bat out of hell and the price performance is really good right I mean that is like that aluminum welded tube race car thing the UM the all flash fast the key point there is just very mature enterprise capability with rich data services I mean you want a exchange administrator to be able to do single mailbox restore coupled into the snapshots all happening automatically all flash fast can do it or or you want an Oracle DBA to be to fire up a whole new Oracle instance and the clones come over and like everything happens I mean all sorts of enterprise data in application integrations that is really the the rich zone for all flash faz and where I see SolidFire I mean what I love about that architecture is the whole way scale model of it 1u boxes connect them in Ethernet back-end I mean it's just it is that Google Facebook style of of super simple just scaling so to summarize it speed for eseries data services for for AFF and and web scaling just easy scaling commodity hardware with SolidFire today what am i close ok well I've actually got the guy in the room I need you're talking about all flash and I think we've had conversations about whether there's an all flash market in the first place and whether we should be talking about all flash as a market now I have such a strong opinion on that but won't finish so that's so are we so it'll be interesting it's obviously one of the SolidFire positioning positions in the market is not necessarily to talk about being an old-fashioned product but to talk about the fact that flash just allows the architecture to do what it does and try and step back from the discussion about about it being fast and you know an old fashioned all the rest of it are you gonna see that work for the rest of the port for to try and get to a position where you're just talking about the product rather than that they're flush particularly flash products so let me make an analogy once upon a time all cars had sticks stick shift and eventually someone invented the automatic transmission and for a little while there was an automatic transmission market and it was kind of a high-end luxury market and you could argue about who had a higher percentage of the automatic transmission market but relatively quickly it just became every car manufacturer that survives has to have automatic transmissions and some of them maybe still make sticks for certain use cases but the reality is you either switch to supporting automatic transmissions or you died like unless you were Morgan or somebody right and so I absolutely think the same thing is happening with flash I think there's there's a long life for disk as a more backup like medium but a relatively shorter life for it at the center of the data center but but this idea that it's an all flash market that was an early nitch thing that was happening before it went wide the same ways automatic transmissions does that hower does that match the strong opinion you have there are storage race and they go fast and they go slow yeah and the parts you use to build them don't define the market the Fast and Slow defines which application it works for you know so calling you know defining an all flash array market you know some other orange company yes can I introduce you to the guys that IDC um yeah yeah don't get me started on Joe Unsworth so um this I promise to be nice this tech field day so anyway I I agree completely it's it's something customers are very interested in it's a change in the market that everybody needs to catch up with but now that we have I mean you wrote a blog in fact about this acquisition perhaps marks the end of the all flash array not as as a thing that people will do because in fact we'll keep shipping them at the end of all flash array as a separate interesting distinct market because that's not even the point so I would recommend Dave's Dave's blog because I think it can't kind of capture that perfectly I'd like to switch to cloud when I look at cloud it's both an enormous opportunity to sit so cool and I'm sure I don't you know public cloud go do stuff on Azure or Amazon or Google I don't need to convince any of you of that it's also very scary to a lot of people the the big issues that I hear on fear are number one is it safe both is it safe for me to get my data back but also who else will be looking at it you know like in the Edward Snowden way is it legal okay you know for HIPAA data or especially European data and the third one will I be locked in so those are the concerns that people have I I got a weird phone call just share this a little bit of a side story I got a phone call asking if I wanted to have lunch with Edward Snowden and I was like number one yes that sounds awesome and number two I don't want to fly to Russia or wherever he is and I actually went and had this lunch who about a dozen of us and Snowden came in the door in one of those little robot things the Roomba with two broomsticks and a monitor attached and he had like a little thing he came up and I Edward snow bot here and I got to ask him a question I've always wanted to ask him I never thought I would get a chance I I said you know from what I understand of you and your situation I'm thinking is it possible that you extracted a whole bunch of data from NetApp gear and he said and he's been very careful about what he'll personally share it's all through newspapers but he said one I won't share that information he said two I will share there was a lot of NetApp gear very nearby me so anyway so that's the downside stuff the the fees fear things but the reality is that it's so enormous ly+ that a lot of people want to try it and so the big question is how can that up help and I would observe that a lot of the issues people have concerns with are about the data part of it right if all you want to do is a bunch of compute cloud is so easy you just turn it on you go get a thousand computers for the weekend and calculate PI or or whatever you want to do it's only the part when you actually have data that you suddenly run into these these issues so net app is well-positioned there's a bunch of things we are doing one of them is how do you back up into the cloud that's an early use case that a lot of people have so we acquired a company called Alta vault that it's basically an appliance you write to it it's got on-board storage but it replicates out to Amazon or actually it uses the s3 protocol so any object store on pram or or cloud a lot of them the number one target is Amazon another thing that we've done is is what we call not at private storage is anybody familiar with with what we've done there now that private storage the idea is this if you're afraid to put your data into the cloud maybe you would like to put your data right next to the cloud so there's hosting places like Equinix and you can get very high bandwidth between Amazon and Equinix it's the Amazon Direct Connect or the azure has the same thing very high speed connect Azure Express route and so you put a storage system that you own it's yours you don't have any questions about what if and all that it's your box but with a very high-speed connection to Amazon so you can have public cloud private storage that's the idea behind net out private storage get a very high-speed link it's a natural place to do replication 2d are just to get started but typically then people can start experimenting so that's a very easy way to get started and finally we also have on tap running as an Amazon you know Amazon virtual machine and you can do a snap mirror from your home environment into cloud on tap running in Amazon it costs five bucks an hour and the fascinating thing about that is like at first it's like you go to Starbucks you can spend five bucks an hour just drinking right I mean it's super cheap which makes people say well how can that even be a business model except if you look at this product it runs at about we've got a fast two thousand we're still calling it the fast 1,000 we you know but it's it's on the order of our lowest end machine and yet if you add five bucks an hour times for your capital lifecycle it turns out you know you better us more than Lisa you're better off buying the car than you are renting it from hertz the beauty of hertz is that you can use it for a while right so it's so so when I look at this cognitive tools how do I see people deploy in cloud and and what's the right model what I think CIOs would like to be able to do is think of cloud as part of the toolkit and tow you know go get started in the cloud and I don't know nine out of ten mighty projects fail right just just go in the cloud but just put a couple rules on run on cloud on tap and then I can bring the data home if I need to for legal reasons again it depends on the type of app you're doing in development whether whether that particular solution makes sense but at the point that that it goes live you might check in with your lawyers you know was that data that legally is allowed to stay out there let me give you an example of a use case I was I was in Vienna Virginia in chatting with a customer and he said he'd been at our insight conference and heard us talking about cloud on tap he had a bunch of web developers and they wanted to move out into Amazon but they were concerned about the data implication how do they keep their arms around the data so he just literally swiped his own personal credit card and expensed it to fire up cloud on tap snap mirrored the data from his bond pram gear into the cloud let the web developers operate in the cloud and then snapped the changes taxes when he wanted right so that's the kind of environment that we're helping people operate let me just close before I get Joe up here with an observation about why I think these ways of helping can work for NetApp and work for customers sometimes when you talk to customers you sort of get this or at least to write the raw razo no VIP you know startups that are doing this stuff or certain types of analysts you get this feeling that everything is going to end up in the cloud maybe just three Amazon Google and Azure and it is a one-way trip and if that's the case then companies like NetApp are just doomed right because it's it's all Amazon Google and and Azure I have a different view my view one of the other tools I like to do for predicting the future is find points in history that feels similar my view is that the best analogy is the transition transition to Windows in the 90s a lot of CIOs in the 90s announced we're going a hundred percent windows in our data center so any of you remember there's a few guys with gray hair any of you remember that money it sound so crazy now I ran some of those benefits that was good consulting money and so it was an enormous migration from UNIX to Windows followed by an enormous migration to Windows to UNIX because everything didn't belong there and if you look 20 years later that's not because Windows wasn't good and it's gotten better and it's a bigger part of the market but Windows joined UNIX didn't kill mainframes or as400 didn't kill mainframes and UNIX didn't kill as/400 and Windows didn't kill UNIX and Linux didn't kill Windows and it's like that new tools join the toolkit and I believe that the the public cloud providers will be enormous Lee important as part of the toolkit but there knock it's not going to be a one-way journey in so whoever can help people keep their arms around their data move it out move it in there's a big advantage there they've a couple yeah so let's do a couple questions and then I'm going to transition to Joe and we'll do questions with Joe and then also questions with J so that even a question it's actually a couple of observations one is I 100% agree first of all I think the future is absolutely hybrid cloud and for a couple reasons one if you think about why people are going to the cloud then stay go to the cloud is all about money and it's that's end of the thing that day it all comes down to money agility - I would yeah but that's also a function of money okay when you really think about it but I was talking last early last year to one of the chief architects at Nasdaq and they had made it they had undergone an effort to move everything into the cloud and they said because it was money and they said but it's a curve right there got to a point where I was way more expensive to keep in the cloud that it wants to bring it back to the local data center and because of all the compliance and other crap that they had to deal with as well as the cost of infrastructure so they actually started bringing some things back strategically so yeah your point about this being able to flow back and forth well what was what was your number you had a percentage of people you've seen bring data back from the cloud was it did you press a button you have to press a button pressing the button now so lots of data on this IDC was actually first to publish probably about a year ago around 2014 and that was already a small majority 51% of all enterprise customers were repatriating data yeah back from the cloud since then there's been North big venture capital partners the 4:51 mary meeker with Katie Ann o kleiner perkins David the state of the internet and so forth all validating the same cycle going on and it's not a pendulum it's a life cycle so data keeps moving around back and forth based on business needs at the time I want to tell a story about this because we had a customer that let nine petabytes pile up in a hyper scalar cloud nine petabytes and they decided that they wanted to bring it home and I just want to tell you about the bill for that first of all the bandwidth that they could get their arms on was required three months to bring the data back and the bill was $600,000 per month to bring the data back now that does not mean it was a bad plan to let it pile up out there it may well have solved an important business problem for them they didn't have room in their data center and they weren't prepared to build a new one like there's no ends of reasons that could have been awesome the real point is you should be very conscious as you let large amounts of data build up and it may well be that even if you want to use cloud resources that at our private storage sitting next to the cloud would be more cost-effective and one other thing I want to mention we have there are our colo data centers next to both an Amazon Cloud and an azure cloud and we can help people do failover from Amazon to Azure where the data is living out of either of them and with that model you're no longer locked into a cloud you can do failover time cut back and forth I don't know I mean they wanted to bring it home but could they have left it there yeah yeah so that was the amount they needed to bring back I don't have more details than that so so let me be really clear now about the structure of what's been going on in this meeting so far so I've been talking mostly about stuff we actually are doing you can run on tap and Amazon's cloud you can do NetApp private storage we can do altar vault backups into Amazon what Joe is going to come up and do is talk about stuff that we're not selling yet that you can't buy yet and the goal of this section is to give you a sense of how we're thinking about what stuff can be coming to add in to what we're already doing right but I just want to be really clear we're transitioning right at this point from stuff that's like real solid you ask one kind of questions - Joel talk a bit about that but - stuff about look this is the stuff in the advanced development lab yes we're planning to ship it but but like as you're putting on your plausibility hat or but wait how do I try that right I just thought it seems only fair to clearly mark that transition I see a question that's a big question so after the EMC merits acquisition whatever you you are becoming the first and most important independent storage vendor but actually so is that awesome or horrible well I don't know that maybe the question is ask me because if you look to all other vendors they are building different kind of infrastructures now I per converged and so that the infrastructure is becoming more transparent okay I just seem to understand so what's your day come on that yeah I would like to push the hyper-converged conversation until after Joe and David both talk but we will have a QA after that the question you're asking let me just acknowledge it is an extremely fair question because the two ways to look at it are wow we're the biggest storage only vendor and to the extent that there's opportunity as a storage vendor to partner more broadly like with Amazon and other folks like that is in Cisco something Cisco that will be well-positioned to do but the flipside is what does everybody else know that we don't perhaps we're doomed right right I mean just in sort of fairness people would ask those two questions so so I just want to acknowledge it's a completely useful and fair question and please help make sure we'd get to it after skip Joe come on up

Channel: Tech Field Day

Views: 5,530

Rating: 4.8620691 out of 5

Keywords: Tech Field Day, Storage Field Day, Storage Field Day 9, SFD9, NetApp, Dave Hitz

Id: Ybg4bSyJ8H8

Length: 32min 41sec (1961 seconds)

Published: Wed Mar 16 2016