00:02
All right, welcome everybody to this pure accelerate talk on expanding the pure E family with flash A AE. My name is Santiago Noon. Uh You'll know me as Tago here and there. Um I'm in the platform product management team and I'm responsible for the pure E family. I wanna jump right into this session. What by uh giving you a quick history of the
00:25
all flash data center? The all flash data center was something that at the time we launched in 2012 sounded a little bit uh crazy, right? We led the charge with the first enterprise all flash array to try and conquer the data pyramid that you see on the right data pyramid is just a quick visualization of, you know, all your data in a data center at the very top,
00:48
you have your lowest volume and most critical highest performance data. That was the first one to be touched uh by all flash storage. And as you move down the pyramid, you have progressively higher volumes of lower and lower performance data. Um going through the different types of media, you know that 15,000 R PM disc uh even struct
01:11
unstructured data in that environment, 10,000 R PM hybrids uh near line discs. And finally, at the very bottom tape, right, with the highest capacity and really most consensi data on it. In 2012, we started capturing experiment from the top down with all flash, all flash made sense for what were really the most mission critical applications,
01:36
the place where performance was worth the premium. But over time, the technology adoption, really all flash becoming more and more prevalent drove the cost down. You made it affordable for more and more workloads, right? So we started trickling down that pyramid to get. First of all those 15,000 R PM disks out at
01:56
that point, the cost of flash made sense against the highest end of disc. So the highest end of disk went away with the introduction of flash blade and direct flash in 2016, we managed to capture not just what was primary storage was typically known as primary storage but also unstructured data in the data center. And so even for unstructured data, there was this concept of critical key unstructured data
02:23
would that would go on all flash systems such as flash blade with the launch of flash ray C fast, fast forwarding into 2019. We went after hybrids but this was big news uh four years ago or so of accelerate when we finally had it with QLC, a cost effective way for flash to replace 10,000 R PM hybrid arrays. This is all those systems that are somewhere in between um mission critical and
02:53
archival, right? Something where you need to be able to access the data, you need it to be uh surely online, but for which it didn't make sense to pay the higher cost of of traditional flash storage. Flash ray C was a solution, flash ray C was the first product to, to be able to address hybrid arrays. And finally, with what we're talking about today, what we have announced in 2023 we are
03:17
finally in a place where we're gonna go after near line disc, the last batch of disc in this pyramid, the only disk remaining before you go into tape. Um and tape has a place in the data center, but this does not. And so with the introduction of E, we're finally at a point where flash makes sense for every workload in your data center.
03:42
Why has hard drive been there? Hard drive has historically been a necessary evil for most customers, right? The cost profile is what made hard drives great. But it came with a catch, it came with actually a few couches in terms of the space hard drive consumed uh the energy uh that hard drives are required to operate and really the complexity and the resources that had to go into managing these incredibly large estate of hard drives.
04:11
If you imagine with some of the uh analysts, data, we've been hearing over and over the explosion of data. Um with A I and everything else, there are projections that this is gonna increase by 10 times over the next few years, right? If hard drives are hard to manage today, they get exponentially worse when that capacity
04:33
increases 10 times. So certainly the tradeoff between cost and uh difficulty is swaying a little bit in the other direction. But at the same time with the E family, we're introducing something that makes cost not an issue anymore. There's a better solution and we're calling that green all flash.
04:54
It is the benefits of all flash and their sustainability and they're delivered at the same price of that as that Neo Neo line disk that customers are used to multiply the two and you get a lower total cost of ownership. Not to mention all the uh qualitative benefits that come with flash, you no longer have to put up with the difficulty of operating a hard drive system just because of the cost. Now you can get flash at the same cost.
05:25
Let's talk about pure E and what's so special about it, right? We just talked about the having the benefits of of green all flash, having better economics than this. These are the key things that make pure E so compelling, right? It is the fact that we have finally found a way to match the acquisition cost,
05:43
not just the long the total cost of ownership, which has been great for certain use cases on flash all along. But the acquisition cost that upfront cost for pure e matches the competitive prices of near line disco rates if the upfront cost matches and you still get the operational benefits of flash and the data reduction that results in much lower long term total cost of ownership.
06:09
And you get that with the best in class user experience that customers are used to with purity and pure storage. Uh And it just turns into a much, much better, way, more efficient, way, more reliable, way, more sustainable way to manage the data growth that's to come over the next few years under the covers. The pure family has two products,
06:34
two products that together span the same set of use cases that we typically see in these repository style hard drive systems, right? We're talking about health care, imagery, talking about media content, libraries and video surveillance, talking about the the colder side of data protection and backup. Uh We're talking about archives of uh log data,
06:58
call data and anything else that doesn't really get accessed on a regular basis, but just needs to be stored for longer term retention, right? Those are the typical use cases we see on hard drive system and those are the same use cases that we are positioning the E family for the family made up again of two products. One we had already announced uh a few months
07:21
ago in flash lady that covers, you know, four petabytes from and above up to tens of petabytes and now new as of this week, flash array E covering the space between one and four petabytes raw um and reading the smaller uh the smaller systems where this data lives, both of them provide the same uh incredible benefits of lower space and power operational cost, lower E waste and significantly higher reliability compared to
07:51
hard drive arrays. And both of them are 100% evergreen looking a little bit deeper into the covers. Here's the flash array E product that we announced this week, right? If you look at the architecture, you can see the common components, the common architecture with the rest of the flash array family, you start with just the base flash array E chassis,
08:17
you grow by adding direct flash modules, E the DFM, specifically the E product family and when needed a storage expansion show, uh you really have the same evergreen architecture we have proven over the last decade. We also build it on the same pure D FA software that we have proved proven over the last decade with the same rich feature set that you're used to on flash array X flash ray C and flash ray
08:42
XL. By keeping the operating system consistent. We make sure that not only are these systems managed the same way and provide a good user experience, but we make sure that we're also leverage all of the innovation we have delivered over a decade and our latest announcement, the flash ray family of unified block and file protocols. Yes, that's available in flash ray,
09:04
E as well on the capacity front flash array is really focused for these colder data use cases. On the slightly smaller end of those capacities between one and four petabytes raw or so, we're leveraging the ultra dense 75 terabyte direct DFME storage modules that were also announced this week and uh with uh three U for the base chassis three U of the expansion shelf. That means between 14 to 48 storage modules between one and four petabytes of raw capacity.
09:36
With the data reduction, customers are used to one flash array that can reach up to 10 petabytes of or of e effective capacity all in just six year of a standard rack with industry leading energy consumption below a watt per terabyte. It is pretty remarkable. The siblings of flash A ray E is flash blade, E again within the same product family,
10:02
within the same use cases. The same general characteristics, just a slightly different architecture that is tuned for the higher scale that is typical of some of these um use cases, right? Not every some use cases are focused on the 1 to 4 petabyte range. Some of them grow to tens of petabytes. And when you start growing to 468,
10:25
10, 2030 petabytes, the scale out architecture of flash lady starts making more sense. So alongside Flasher ray, we have a scale out architecture and flash blady again, future proof. Again, evergreen again, leveraging all the similarities to the to the rest of the portfolio um that we've had for a long time. So the same pure DFB operating system in this
10:47
case, unified Filin object protocols instead of file and block as was the case on the flash Ara E side, the architecture itself on flash blade, E is focused on efficiency similar to what we talked about in flash rate. E in this case, we have two types of blades. We have a EC blade that provides both uh processing power as well as storage. This is where all of the operations,
11:15
all of the management, all of the IO gets taken care of uh plus a number of a number of expansion uh chassis with ex blades. These are storage only blades that only provide additional storage capacity. So by having this ratio of little compute little control power to lots of storage, we're really fine tuning the flash blade architecture for the use cases where cost
11:42
and not performance are the key metrics. Here we start where flash array left off flash array left off at about four petabytes. Flash blade E starts at about four petabytes with 40 times 48 terabyte direct flash module E today per chassis. Um And later this year, we will also be using the new ultra dense 75 terabyte modules for
12:05
three petabytes in a single five U standard Rata chassis again with less than one watt per terabyte of energy consumption. This is absolutely misleading, not just against hard drive arrays but against any type of storage array. We've talked about the 75 terabyte dfmes on each one of these products. These are also brand new as of this week for a pure storage announcement,
12:32
they are the largest capacity storage modules in the industry, 75 terabytes of QLC in a single direct flash module. It is pretty crazy to think how far we've come. Um It is one of the key things that allows us to be competitive with hard drive. From a cost perspective. The density, the efficiency that putting so
12:56
much storage into a single module brings us is all the efficiency that we pass on to customers in the form of a cost that's competitive with hard drives. And the key to that lower cost is really the space savings that we get the energy savings that come with such density and the reliability advantages that come with this density three things that I'm gonna talk in more detail in just a little bit.
13:19
But really if you want to take away one thing that enables us to do what we're doing with a E family, it is direct flash, it is these extremely dense direct flash modules that are unique in the industry. And key to the advantage that that we're showing here to summarize under the covers within the pure E family two products, flash array E and flash blady.
13:44
Both for the same use cases. The main difference, if both have the same incredible economics, the main difference being on two fronts. One, how much capacity uh are you aiming for the smaller side of this 1 to 4 petabytes for all focusing on the flash arrays E scale up architecture, the larger capacities from four to tens of petabytes raw that more um tuned to the flash blade E scale out architecture.
14:11
And on the protocol side, you unify file and block on the flash ray side, unified violent object on the flash blady side. And so two siblings within the same family delivering the same incredible economics that are competitive with disc. So now that we got the cost question out of the way cost is no longer an obstacle.
14:31
We said that the reason hard prices were so prevalent in data center was cost without no longer being an obstacle. What are the benefits? What are the reasons that we think flash is absolutely a no brainer when compared to this. First of all, it's efficient efficiency, flash array and flash blade E the E family as a whole is just absolutely more efficient than the same capacity delivered in the form of hard drives.
14:56
And this goes back to the revolutionary um technology that we have in the red flash incredible space and power savings, very easy to understand. Right. If I take every single drive, every single hard drive that is typically 10 terabytes on. So or so in size for these den uh dense systems and replace them with a 75 terabyte direct flash module.
15:20
Just that seven times increase in density, you can see the effect that has on the system level. Right? It means a lot less compute power to support that uh storage, it's a lot fewer racks to support that storage. The moment you make every module seven times the density. Yes, there's a little bit of um snapping to the boundaries of a chassis or a system which,
15:42
but at the end of the day, you do end up with just 20% this space on power compared to a hard drive rate. It's pretty incredible with more efficiency also comes more reliability, right? Reliability is, is one of the um worst parts quite honestly about running hard drives. Those uh hard drives fail at about twice the rate compared to industry standard S sds.
16:09
And some of that just has to do with the fact that they have moving parts. We all know that uh in every single system, be it. You know, my car and enterprise storage, whatever it is, the moving parts are some of the first things to fail. And yes S sds don't have moving parts. So their failure rate compared to hard drives,
16:26
I is about half during and that's during the typical operation period without even worrying about the uh long tail of uh failures once the drives get old. But compared to S SDS, we all know that direct flash modules are even uh more reliable, right, direct flash because we talk directly to the flash because we treat it at the global level because we don't have to increase the amount of D ram the size of the controller as
16:56
we make these drives denser and denser. Not only are we getting the advantages compared to hard drives, but we're also getting the seven times advantage compared to um S sds as well. Combine combine the, the no moving parts advantage with the direct flash advantage. And with what I just talked in the previous page, the fact that we have seven times fewer
17:19
drives because of that incredibly high density multiply all that together and you really get quite easily to more than 20 times fewer failures 20 times. Think about what that means to your schedule, to the number of times that somebody on your team has to go to a data center or you have to pay somebody to replace drives and components as they fail. 20 times a liability means turning a failure
17:48
that we used to have on a weekly basis into something that you're gonna worry about once or twice a year. Right? It's really your time coming back to you and finally, sustainability. Let's talk about sustainability, right? We've the cost is one aspect of it.
18:07
The power savings is one aspect of it. Well, let's talk about the environment as well and just translate what these metrics mean for the environment, right? We talk about 1/5 this phase, we talk about seven times fewer drives and 20 times the right of liability. You combine all of that with the fact that not only do we fail less during the normal
18:27
operation of a system, but we also have direct flash modules that last significantly longer. All of that combined means 85% less e waste compared to running a hard drive array have components that have to be replaced, less often, way fewer components to replace in the first place and way less just pounds of material to have on your floor. Right? It's fairly straightforward.
18:53
It is just a more sustainable product for the environment and all of this just keeps getting better, right? I talked about direct flash being the enabler for what we're doing here. But look at the road map for direct flash, look at what we know we can do over the next couple of years in terms of drive density drive density that we have announced this week at 75
19:17
terabytes. We know this is gonna get to 100 and 50 to 300 terabytes just over the next few years. And with the density improvements that we keep making on the red flash come proportional savings on the space power cooling cost, re reliability, sustainability to you, right? All of this just keeps getting better and it's all thanks to our unique direct flash technology.
19:41
If you want to hear more about this, please tune into uh the direct flash technology, deep dive talk with our chief architects. They have a lot to uh share with you about this. It is really cool stuff. So we thought I'm gonna leave it at just one statement, flash and flash lady together form the pure E family and the pure E family with a co acquisition cost that is comparable
20:09
to disc and incredible and very clear advantages over the operating life, make it a complete no brainer to replace it, right? So with E here, we know that the end of hard drives is here too. Thank you for tuning in here.
