Pure Storage’s Justin Emerson on Analytics Performance and Flash Storage

hi I'm James Maguire and on today's e speaks we're talking about optimizing the data analytics practice in the Enterprise including the role of high-speed data storage in boosting analytics practice to discuss that I'm joined by an industry expert with me is Justin Emerson principal product manager and Technical evangelist for Pure Storage Justin happy Friday to you and very good to have you with us today happy Friday thanks for having me James really appreciate it so data analytics more important than ever before you know companies can no longer run their business on on gut instinct anymore and yet still companies are struggling at in terms of getting the most from their analytics practice what's a major challenge that companies face with with data analytics in your view for all the effort and expense what still holds companies back well I think you what's interesting is is that when we talk about analytics um oftentimes that first word that you said data gets dropped off of it and it used to be that we talked about this in the context of big data which sort of put the the data out front there and kind of talked about it's about the data and now we're talking a lot more in general about the process of analyzing and what tools are we using to do the analysis and how we're starting to use artificial intelligence and machine learning to do the analysis or or various things like that but really at the end of the day all of that is based around data and so in the context of analytics whether it's about you know observability or about diagnosability or about security or all or or all these other things right really it comes down to collecting good data and making that data bu accessible and one of the challenges that customers have found is that you know ever since we've all read that you know that old cover of data is the new oil we've been just sort of collecting everything like hoarders um but we've never really had good processes to do stuff with all of that data and so we've ended up with data that either we don't have indexed or don't have cataloged or data that is in lots of different places and are in different silos and so I think one of the biggest challenges around data analytics today is understanding the scope of all of that data about data governance and about understanding what you have and then being able to actually do things effectively on that it's it's almost a bigger challenge to know what you have before you can even get to the what can I do with it yeah I think it's amazing for companies and I think they the analytics practices have sort of grown up P Mel that there are a lot of buckets and silos and it's we've got sing in the Tulsa office we've got sing in the Topeka office different formats I sometimes I wonder you know with all that spread out data how companies get anything done yeah and what's what's interesting is is that because the problems that a lot of these analytics um uh programs have run into is that a lot of them have started to undertake um early on they all sort of De decided that we need to sort of manage storage ourselves because the storage problems that we're dealing with the data problems we're dealing with are so large that most of the other uh organs within the the organization are not capable of handling them and so a lot of what the development around big data and then analytics has been around is just collecting and managing data and that has sort of happened as a complete separate silo or a complete separate um uh you know uh configuration versus what standard it might have been doing so what you end up with is analytics packages or pipelines Etc that are um totally uh siloed off from maybe the central uh you know storage or Central data repository or the data strategy that an organization has and then if you have multiples of these packages maybe so you know something like a like a message bus then putting into an analytics or log Analytics tool and these things chained together each of them may have its own storage Silo because each of those packages has grown up with its own storage backend because um they were developed in a time where they could not rely on a shared storage service to be able to deliver the kind of scale and the kind of performance that they needed so the time in which all of these tools have evolved has sort of exacerbated that problem of data silos so that's the that's the problem and clearly there's there's a number of problems there how can companies solve these kind of issues is there industry momentum toward any any solutions yeah so I think what's what's what's a little bit funny is that um you know we're we're start we're rediscovering the same lessons that we've learned again and again every decade of of it which is I like to say that things happen very cyclic you know in a very cyclical way like we we go from let's centralize everything to now let's distribute everything now let's centralize everything you know it was like vdi was just time sharing on Main frames you know just removed by 30 years um you know we go from parallel buses to serial buses and then back to parallel buses then back to serial buses and um and of course Cloud itself helped you know move move data data all over the place absolutely right cloud is just you know centralizing it on somebody else's computer um and so uh I think that you know this storage has gone through this same thing over and over again which is you know we we bring storage together and then uh you know as one example oh we're going to we're going to break it apart and distribute it out because uh we're doing so much and we need all this you know data close to where it's being analyzed and and I think we're we're we've started over the last five years to hit that sort of cycle going back the other direction which is that one of the big trends is that in these kinds of um or of uh processes we're starting to see a Recon convergence of storage and so whereas many of these packages would have in the past past been focused on okay now I'm going to distribute my storage out to all of my nodes that are doing the processing we're starting to see a recentralization of all of that storage and so that carries with it huge benefits which again everyone has already figured out years ago in different contexts um but because of the scope of it it sort of necessitated this well we can't do everything in one place anymore let's distribute it out right it doesn't we can't put everybody on the main frame let's put a PC on everybody's desk so um that trend has um really accelerated with what you brought up the cloud um because as people say move some of these workloads into the cloud it becomes really really financially in um advantageous to centralized storage because that means you're using less of it and therefore you're paying your cloud provider less money per month and so that's highly incentivized uh some of these uh packages to evolve in ways that can leverage centralized storage and that work and that that uh evolution is now feeding back into on-prem infrastructures as well and so I think one of those big um solving Trends is is bringing storage back together which not only helps to um make it more cost- effective but also break down those silos that we were talking about which is now I can have different stages of that pipeline accessing the same data instead of each of them keeping a discrete copy of it uhuh okay makes sense well let's drill down to to to the Pure Storage Solutions I mean how does Pure Storage support the analytics practices of of its clients I mean what's the advantage of Pure Storage right well so pure is a uh you know an Enterprise uh and cloud storage company and we have lots of different solutions for customers across all sorts of different vertical and applications but in the context of analytics right one of our our biggest strengths is that we can provide a highly scalable all flash unstructured data platform which customers can leverage for many many different different applications not just within the analytics space but with others but in the in context of analytics we can uh offer uh you know highly performant highly scalable storage over you know whether it's a file protocol or whether it's an object protocol and that's important because a lot of applications are in a middle of an evolution going from one to the other so you may see people using file storage for for certain parts of their workflow using object storage for others but over time some of these are going to move more towards object and so having a really flexible storage platform that can do all of those things is really important because it doesn't uh lock a customer into a particular methodology or or particular architecture it gives them the flexibility to do that and as you as you bring all of these you know different tools together to run on some kind of centralized storage um the scalability and the performance of that storage becomes enormously important because you don't want to you know you want want it to grow with its capacity and performance along with your business and if your business is growing you know in in in an ideal world that means you're you're being more successful you're probably collecting more data you want to do more analysis you want to be more effective with that analysis and bring on new and more interesting tools and those may require even more performance you know one of those reasons that people uh spread all that storage out to the edge in the beginning was that the centralized storage repositories that they could have leveraged would have crumbled under the weight of all of the data and all of the performance that that data required in order to do the kinds of processing that we're talking about so um pure being an all flash storage company we do flash better than anybody else in my humble opinion um and uh that means that we can build really really intelligent really really scalable platforms for these analytics tools and almost just as important they're they're easy to manage and for these applications where the storage management has for the you know recent past been a burden on the application team because they've not been able to rely on a centralized storage provider um for them who are generally speaking not natively storage people having a complicated storage engine that they have to manage is probably very low on their list of priorities so having something that just sort of works in visibly to support their workloads is super important yeah what I'm hearing you say really is if you're going to gather your your data all in one place you definitely need to be supporting that with a with a a storage system that can really handle that absolutely and and one that uh doesn't just handle it on day one but can handle it on into the future be evolvable be scalable uh continually improve over time and that's something that we really pride ourselves on delivering storage platforms that get better as they age not not uh are obsolete when you take them out of the box you know this there's one other I think very important topic that it's related it's uh it's ESG environmental social and governance and and and how does that relate with the issue of data storage with with the shift from from hardest to flash what are the implications for issues like power use and sustainability that's an absolutely great question and it's top of mind for so many different people it it tends to vary from geography to geography like we we find this is a much um more I mean it's it's like a board level issue in many companies and say for example for in Amia but just more Broad in general um you know ESG and and um sustainability are becoming higher and large bigger and bigger priorities um for companies and I believe uh the statistic is something like 2% of uh energy consumption globally is consumed by data centers so that's that's a non-trivial number um and our estimates are that storage is something like maybe half a percent uh so so it's about a quarter of that overall power uh energy consumption so so um as companies store more and more data as companies process more and more data as as you know analytics tends to to to feed more and more in it's sort of it's we are be we are um it behooves us to really focus on how do we do that efficiently how do we do that sustainably um just you know not only for business reasons but also for for social and ethical ones so go ahead please yes yeah so so hard hard diss you know are a technology that have been around uh long before I was born and uh uh they uh have uh done a very good job of being the sort of primary online storage me mechanism for you know for o over 50 years so it really behooves us to think about how do we do this sustainably and you know hard drives have been around since long before I was born um but they and they've done a very good job in the last 50 years of being the you know primary online storage medium for um for organizations but flash is just fundamentally so much better the only reason that companies have not moved every single bit of data over to flash has been cost um and so one of the big advantages of our approach to doing flash is that we are able to deliver solutions that are so much more power efficient and cost effective uh than than other Solutions on the market that we're now looking at how do we replace disk across the board uh in organizations and when we talk about analytics there's so much data um you know being able to make a dent in the the you know sustainability and the efficiency of that data platform can really move the needle for organizations I was just at a another uh company earlier this week talking to them about one of their Solutions and they could save millions of dollars in power over the next 10 years just by moving some of their workloads from disc to flash so even from a dollars and cense standpoint it makes sense but also from a environmentalism and an ethical standpoint it makes makes all the sense in the world on top of which flash is just a better medium for uh for these analytics workloads so many of them require the kinds of you know IO and the kinds of random um uh performance that dis does a very poor job of providing and so flash is just better across the board and and as I said earlier I think we do flash better than anybody else but you know I'm biased um so I think that if we think about when we're talking about analytics we're talking about any really large impactful source of data um every organization should be thinking about how do I do that efficiently how do I do that sustainably and pure is in the Vanguard of companies in our space talking about how our Solutions impact that so we released a whole ESG report um uh about a month ago um where we showed all of the benefits of our Solutions compared to others both all flash Solutions and dis Solutions um and what a difference that can make for organizations in terms of sustainability so I really do hope um that more people are examining those things um when they're making buying decisions or when they're making architectural decisions um because if it can make you know a a dent in overall Global energy consumption I think that's just a win for everybody across the board so so flash is good for business Al good for the planet that's a great combination it seems like then that an all flash platform is essentially inevitable in the Enterprise at some point in time absolutely in fact our uh our CEO at our accelerate conference said that um you know we we think that there will be no more hard drives being sold by 2028 um which is a pretty I like to call it a hot take uh it's a it's a pretty it's a pretty bold statement I'm sure that there are some people who uh would disagree um but you know I think we are very confident in um that uh like like many technology transitions before right that uh you know we're not using eight tracks and we're not using uh I would say vinyl but that's not it's made a comeback so you got to be careful about your metaphors um but uh you know we're not using the horse and carriage uh anymore uh for for daily transportation and I think that uh by the end of the decade we're definitely not going to be using spinning magnetic platters for our storage anymore there's just a fundamentally better way to do that and the economics for that aren't being driven by the data center they're being driven by all of the devices in our pockets all of the devices in our uh all of the electronics in our our cars and our televisions and everything like that so um I think the the the Arc of history is sort of bending in that direction inevitably clearly Justin I think you said it a lot of good stuff I I certainly learned a lot uh thank you so much for sharing your expertise today and please come back and talk with us again sometime oh James it' be my pleasure thank you so much been great great been great being on

This transcript was generated automatically from the video's captions and may contain errors.