00:00
Thank you so much for being here and I hope everyone had a great day at Accelerate. We are here to talk about not this. Yes. Every why every use case deserves the right story a solution. Now at pure, we are incredibly proud of our robust portfolio of solution from our flash array flash play port works and Evergreen family.
00:26
And we serve a wide variety of customers and use cases. It's not atypical for us to hear about from customers who started with one workload, one product and from then expand into a variety of other workloads and product line as well. So today during this session, we will have not only a quick overview about our flash array and flash blade product line, provide a quick overview of some of the use cases we serve.
00:58
But more importantly, we'll spend a lot of time talking about from a very special customer who has gone through this experience of starting with a single solution and expanding from there. So we'll have plenty of time at the end for Q and A. So if you have questions, we'll absolutely take them at the end. And we, we we are happy to stay back a little later and answer any questions as well.
01:21
So with that, let's get started. So I'm Bishu MSRA director of product marketing for the flash plate product line and I have with me, I'm Manisha. We, I'm senior director of product management for flash file services. Thanks Manish. So let's start off with like,
01:42
you know, we talked about a wide variety of workloads. When we go in and speak to customers, we hear about the breadth of their workload requirement. They can be some of the block based workload requirement, think mission critical application databases or they can be file and object based workloads. Think analytics, think artificial intelligence, machine learning,
02:08
think even the backup environment, the list is pretty long. But one thing is really common when what we hear from the customer, the challenges across the workloads are really consistent. You have more data that growth in these applications has really been driving an exponential growth in data and really managing that as an incredible challenge supporting
02:34
these workloads require storage architectures which gradually continue to become more and more complex. So the management becomes an incredible challenge. The third one is around the Opex challenges that's mostly around power space and cooling, the require solutions which make it easier to deliver the storage on a smaller footprint in a more power efficient manner.
03:02
The fourth one is around the reliability of the data and the ransomware protection. So it's really essential that your solution is not only available but it can recover from the threat of ransomware attack if you are there in that unfortunate situation. But the last and probably just the most important one is the ability of scaling your storage environment based on whatever the future is going to throw at you.
03:32
And that means scaling your performance, scaling your capacity and protecting your investments in storage that you have today. That doesn't, it doesn't matter which workload you are talking about. That's something that we hear consistently from customers. And our entire flash array and flash product line really delivers on these elements.
03:54
And these are some of the core tenants by which we build our product on. So what our family looks like is our in case, by the way, quick show of hands, how many of you are familiar with our flash array product line? Wow. OK. Uh the flash blade product line, all right. So slightly fewer people. So a quick recap.
04:16
So our flash a product line is a scale up architecture, unified block and file services right now. Some great announcements today that you know, there are plenty of great sessions. I would highly encourage everyone to check them out. Some amazing is there you have the flash array X for mission critical workloads,
04:36
very consistent performance and for virtualized environment you have and then you have the flash array XL. This is for like you know, applications which require really extreme performance really high scale. You have this flash array C which is for like you know, lesser demanding applications, more balance of performance and capacity.
04:59
And then we have the flash blade family which includes the flash blade S and S 204 100 which delivers a scale out architecture with unified file and object storage capability. It provides the right balance of the most extreme performance and more capacity optimization. But I am saving probably the new exciting news for the last the pure storage E family, which includes both the flash blade E and the
05:27
flash array E which was announced today, which is targeted towards the repository workloads which historically have remained on this way storage till now to gather this entire portfolio really delivers on a broad range of all of the use cases that we were talking about. Irrespective of what the protocol is, what your performance requirement is, what your economic requirement is.
05:53
But there is a reason why flash array and flash blade really deliver on this. If you heard the early morning keynote, there are some elements that we have really designed to make sure the portfolio can meet the requirements in this manner. The first starts with our commitment to direct flash technology, our direct flash module which we build, give us a tremendous advantage over not only spinning
06:23
this but SS DS which pretty much everyone else in the industry is using right now. They are more efficient both from the density performance and efficient power efficiency perspective. Add that to our world class data reduction capability team that you have a really compelling solution. Our operating system across flash array and flash blade is the same purity operating system
06:50
which provides a seamless experience for our customer. And we also deliver world class energy cap energy savings and ESG savings across both of our portfolio. Lot more details around this. So you know, if happy to talk more about it after the session or we can find us in and up booth, we can get into that the other element
07:14
which is our pure one as the application and your environment is growing. Really being able to manage and monitor your fleet in a seamless manner is a big unique differentiation and save probably the best for the last a big unique differentiation which none of the competitors can really match us on is our evergreen capability now because we have developed our hardware and software from scratch in a manner
07:45
that the system always remains future proof that any tech debt that you have as the environment grows, we can upgrade our hardware and software seamlessly to catch up on that tech debt, you protect your investment, you can upgrade your controllers, you, your support cost remains flat and fair. Evergreen forever is a incredible differentiation in the market.
08:12
Add the flexibility. What evergreen flex and evergreen one provides to consume the storage in a more uh cloud consumption model that becomes a really powerful tool for us, which is available across the portfolio. But to provide a quick overview of the use cases, let me hand it off to. So as bish mentioned, if you look at the workloads on the top,
08:40
uh you know, there is generally class of workloads called structured workloads. They align very well with the block protocol axis, but it's the final object that has a very very wide variety of different workloads. And by that nature, with the needs of those workers. So this is roughly how we would position two different product lines.
09:02
You know, flash obviously is a scale up architecture and it has low latency and you can scale in terms of capacity and that aligns very well with virtualization home directory, those kind of use cases, flash blade on the other end is a scale out platform. It gives you really large scale capacity as well as you know performance scaling. So those workloads that align better with that are the ones that we would position flash blade
09:32
for. And then there are some in the middle where it's really the workload itself has a very wide split. So depending on your scale needs you position one or the other. However, you know, you can actually use a combination of flash array and flash blade for each of these use cases because what we showed on the previous slide was what you would
09:52
use for the primary copy of the data. But you know, usually that's not the end of the story, the data has other needs outside of the primary copy. So I'll go through a few examples where we'll look at, you know how you can use both flash array and flash blade together in a single architecture to solve an end to end use case for for specific workloads.
10:13
So starting with enterprise imaging, this is a simplified view. This is more of a data storage view, obviously that ecosystem is a much, much more complicated ecosystem. So you know, on the left, we've basically simplified that into if you look at it from the storage standpoint, there's a database, there's VM and operating systems running in there,
10:33
there's a short term image cache and then there is a longer term data repository for the, you know, keeping the images for a longer period of time. Roughly speaking, the first three account for about 30% of the capacity needs and bulk of the capacity needs are in that longer term, you know, storage repository. So you could map that 30% you could use either flash array depending on your performance and
10:59
capacity needs or go with the flash blade S and then the E family would be a great fit for that longer term you know repository. Now, what we hear from customers before and after using PR is on the right side of the slide where you know some of the key challenges that customer, you know, when they come to pr that they are facing, one is there's really a large data
11:24
center footprint that they're trying to reduce. There's energy and space consumption that they are grappling with. And there is forklift upgrades periodically required when they refresh the the ecosystem or the infrastructure that they are facing with, as well as you know, nickel and diming on the features. So that's, you know, roughly the problem that most customers would come to us with after they
11:49
moved to Apr all flash data center, the benefits that they would see is a very lean infrastructure dense compact. Many times we'd see several racks being consolidated to a few rack units of flash storage that's not uncommon to see. Then obviously green B talked about that there is simplicity angle. So you can, although these are two separate hardware architectures and they've got
12:17
different properties in terms of how they scale you can combine them from a management plan standpoint using our pr one SAS platforms. And that obviously gives you an end to end sort of solution for that use case. The next one I wanted to talk about is how do you build a complete data protection solution? Starting with the primary copy of the data, having local snapshots.
12:43
Uh These are the protection services that we would uh you know, consider as part of our overall data protection solution then having a primary data protection site your dr site and then an optional data only bunker. And either of these can be done with flash array as well as with flash blade and a combination of two for different use cases. On the data center side,
13:06
you may have, you know your primary copy on flash array. You may have a second data center where you are mirroring data to maybe even a synchronous mirror. You may or may not have that, then you have local snapshots which either of the product lines, you know support. There is one important element which is around safe mode.
13:26
That is more and more we heard from customers where you know, ransomware as ransomware attacks become more and more common. That's a common service that we've got across both of those product lines. Then you can go to a second sort of dr site or we call it data center three in this particular image. And finally, the, you know, with uh you know
13:48
flash blade uh as well as flash array, you can have a data bunker. So you can have an end to end uh data center wide data protection uh solution with either of these products. The same thing for here. Basically, we look at a combination of flash array and flash blade. So if you've got a database,
14:09
say it's an oracle instance that you're running, it could be virtualized, it could be a bare metal instance, it could be rack that would typically map onto an X. If it's not that, you know, performance oriented, it could also reside on a flash array C it might even be on a third party, you know storage solution, the primary copy, then you, let's say you're using a application level data protection
14:32
capability like R man. So you would again then have two different options on the where you store your data protection copies. So you could have a block or a NFS option with your Flasher AC, you could obviously use flash blade to host those backup copies depending on what your scale needs are.
14:53
And you could use NFDNF or object. And this is a very common combination that we see our customers use. So the last one that I'll walk you through is a plunk smart store use case. This is basically got two different tiers, the primary tier, maybe if you've got you're using containers, you do port works along with flash ara.
15:20
So that would store your, you know, your ingest or the cache layer. And then the warm of the remote tier would be something like a flash blade. So that would basically a combination of these two being used together to solve a end to end use case. So those are some of the examples. And now I wanted to invite Eric Andrews who is director of it for to share his
15:47
experiences and you know, basically just his journey with Pr All right, Eric, welcome. Thank you. To be here for being here. All right. Um, Eric, let's start with a softball. Um, tell us a little bit about and like, you know, your role and your team's role.
16:14
Sure. So, uh, uh, I'm the director of it and our, um, our job is to manage all the infrastructure. So that includes, um, we operate out of eight data centers. So we operate the, uh, the compute the network, the storage. Um, Our team basically is responsible for everything up to the operating system that runs
16:35
on all of our, our infrastructure and, and today we don't use any public cloud. So this is all our own uh hosted infrastructure. Um Nick, let me speak to that real quick. Nick, we are a traditionally, we've been known as a UCAS provider, hosted voice over IP services is what we offer. Um you know, as a company that we started transition to a,
16:58
a provider and we integrate mobile technology into our, our voice service. We have MS messaging. We've implemented some social technologies into that call center capabilities, customer analytics, interaction, customer interaction technology to look at the customer's data, their calls, their messages and things like that and provide
17:22
predictive analysis back to them on their customer base and things like that. Awesome. So fair to say your team kept keeps next. We do. Yeah. Um You know, we make sure that uh our customers uh you know, have service stays up. So I wanted to get into a little bit around. Uh, you know,
17:42
obviously given that you're a UCA provider, downtime and outages are really impactful for your customers. So how is your team helping your customers actually make sure everything is running all the time? Yeah. You know, downtime for us is really important. Um, you know, I think that as I can make a comparison real quick that if customers or
18:05
customers of like cell phone providers, you know, t Mobile at T Verizon, your cell phone, if you get a drop call, you know, what do you do? You just dial the person again, right? You're probably not opening a support case with Verizon saying I had a drop call. But, um you know, with VoIP services, customers are very finicky about the service.
18:22
So if you have one dropped call or you pick up that handset and you don't have dial tone, you know, that customer can't, they can't do business or they're missing calls from their customers. So about eight years ago, our business, you know, the business decided to say we need to be able to provide 100% up time on core dial tone service. Basically, meaning that when a customer picks
18:45
up a phone or tries to dial a number or tries to get or receives an inbound call that it's completed. Um And so, you know, we took that challenge on and uh you know, like I said, eight years ago, we decided to heavily invest in infrastructure. We built out at the time, we grew from two data centers to eight. we invested in pure storage, we use Cisco U CS the flash stack model.
19:07
Um And we really put all in on redundant infrastructure to ensure that, you know, we can prevent customer outages as much as possible. Can you talk a little bit about the role of storage in that particular aspect? Yes. So um we uh you know, we operate uh uh oracle, we have an oracle rack database system that you know holds our holds,
19:32
you know, for storage, uh customers call recordings um as well as a big component of of is, is how we build is based on rated phones, right? So every phone call that comes through our system, you know, regardless of whether a customer pays per minute, we have to rate the call so that we can, you know, we can then build a customer for the calls.
19:51
And so all of that storage, all of that data gets stored in our storage platforms today. So, I mean, I can imagine like, you know, that being a huge competitive advantage to you. But like I can see also like, you know, your team playing a crucial role in your customers having the a modern business communication platform just staying on the storage angle.
20:19
What kind of storage challenges did you face? So, you know, like I mentioned, we traditionally were we were a U CS provider. But as we've been, you know, evolving into more of a provider with all these other applications as part of our suite um storage. That's where we've really seen our, our, our storage growth and storage needs increase.
20:39
And you know, because we keep track of customer interactions and um you know, we, we're, we're doing a I predictive analysis on like customers, phone calls and their call recordings, their S MS messages so that they can then see if their customers are happy with the service they're providing. So, you know, storage pretty in in our case, in previously storage kind of just like it,
21:02
you know, grew at a nice even pace with customer acquisition. But as we see now the more customers use our platform, it exponentially grows and we're seeing a pretty massive increase in storage growth and need for performance in our storage platforms. So, um let's step back a little. Can you give us an overview of the pr deployment in your?
21:26
Sure. So um we have uh today we have eight flash arrays and two flash blades, um two flash array seas. Uh And we are also currently actually looking at port works um for some of our databases and service um uh system. So, and how long have you been APR? Yeah, so we, we purchased our first uh pure in 2014.
21:53
And um yeah, so we've been with pure for, for quite a while uh at the time they were uh they were fa 4 20 models. Yeah. Um And that was purchased originally for an Oracle database system. Um You know, we weren't as a U CS provider. A lot of our systems still require bare metal because of um uh code processing and things like that.
22:17
So our, our soft switch platform hadn't really certified for virtualization yet. Um But then within the last five years, they started certifying us to be able to run some of that uh in virtualization. And you know, I went to the business and I said, hey, we need to move our virtual work workloads as we build these out on pure. Um We started building them out on that same pair of fa 4 20.
22:38
Uh And that was, that was great. Um You know, it handled the oracle workload and the, the virtual workload is just fine, uh you know, fast forward a few years and it was time for our first Evergreen upgrade and then we upgraded that to an M series. Um We purchased some additional arrays as we grew to additional data centers.
22:58
Um And then that then progressed to an X uh Evergreen upgrade um and purchased some additional arrays and yet, like I said today, you know, there's eight array, eight X arrays, two C arrays and two flash blades. Eric. I know we have talked about your experience before pure long hours in the data centers. I think I would love for you to share that with
23:23
the broader uh folks here about what that looked like, you know, for like the an evergreen upgrade. Well, no, I mean, your experience before pure, what does your storage experience? So, um you know, I have my other storage engineer here in, in the room but um uh you know, we've spent many of long nights and calls with other
23:40
storage vendors in outages in downtime and um you know, moving to pure uh definitely took, uh you know, it took a, the, the complexity of managing those, those other array systems and uh and, and the maintenance and the upkeep and this warranty and support renewals to ensure that you can get spinning disk drives. All the, you know,
24:03
and every the simplicity is, is just hands down amazing. I mean, I we've, we've probably gone from, you know, 10% of the time we spend on storage arrays from what we used to spend uh before. So you, you mean like what we discussed during the keynote that we got the nights and the weekends back we have? Yes. Ok. Yeah, that's good.
24:22
Uh Exactly. So as is transitioning to like the SAS business model and requires like, you know, the more efficient, reliable storage. How much of a difference has pr made on that front? Uh You know, that's where I, I, like I mentioned before that, you know, as we see the performance and growth needs exponentially grow,
24:43
um what I found is that to, to, you know, upgrade our pr A S or add additional storage is really simple compared to other storage platforms. You know, we've worked with, um, you know, sometimes we'll wait for an evergreen upgrade to occur. Other times we'll, you know, advance that evergreen upgrade, you know, for, you know, a minimal fee.
25:06
You can, you can, you know, get your evergreen upgrade a little early. We, we've done that a couple of times when needed, you know, adding data packs to arrays has been very simple. Um, you know, and I know you can swap data packs between arrays. So if we have one site that isn't using it, we can migrate it to another site,
25:21
things like that. So I think peers just allowed us to really, uh you know, to ramp up quickly. Um So I, and I think in uh in uh some of our earlier conversations you mentioned, like, you know, you're looking into the cloud like, you know, what drove you to the cloud and like, you know how we,
25:39
we attempted to go to the public cloud um in 2019 until about three months in, we saw bills in the, you know, six figure mark. And the first thing we did is we were like, ok, we can do this, you know, we can do this better, cheaper on our own and with less outage, you know, um we're pulling out and so we've made an effort,
26:04
unfortunately, going in was really quick, but coming back out takes a lot longer. Um Right. So, uh we, you know, over the last two years, we've really made a big effort to pull everything out of public cloud that we can and invest in our own data centers. Um So we had the infrastructure footprint already. So we, it was easy, just build up on those
26:24
blocks and, and grow it. I mean, we hear that a lot from customers that have done stuff like, you know, that cloud isn't always the right answer in every scenario I think is a great example of that. So, Eric, we've talked about your um you know, how you started your journey, can you talk about the use cases that you use for flash
26:42
array and flash? Sure. So as I mentioned, um you know, we started with flash array for um oracle rack database. Um and just to speak to that real quick, we went from, you know, maxing out our previous, you know, storage array at the time with that Oracle database. It was like 90% utilization to when we replace
27:03
it with uh you know, flash and and pure went down to like 10% utilization. It was a huge performance increase. We're like, what do we do with all this, you know now. Um so you, you know, we, we deployed oracle on that then uh you know, a few years later we started moving virtual workloads over to it um added additional arrays.
27:21
Um And then, you know, the business decided to attempt something different. And um a few years back, we decided to uh deploy open stack. So we run uh open open stack. Our team runs COTIS uh within uh pure flash array. Um We then decided that it was time to really replace a lot of our spinning disc. And we needed a solution for call recording
27:48
storage. And that's where we came along and started using the flash array C. So we added the flash ray C for our large workloads more. You know, it's nice to hear that the E has come out. That's uh that's really fun. But the C basically holds our holds our data that sits at rest,
28:03
you know, longer the before the performance isn't as necessary. Um And then fast forward to just even a couple of years ago, we uh we needed a solution for a component of open stack. But as well as uh Splunk, we run Splunk in our environment and we needed a solution for the warm storage component of Splunk. Um And that's where we decided to go for with
28:28
flash blade for that purpose. So we run uh today we run on flash blade, we run uh Splunk smart store. Um And also we do uh there's some components of open stack, we use NFS for there, but also recently we've started moving some of our elastic search workloads over to uh flash blade. Um And we've had some good luck with that.
28:50
Uh you know, as of yet, we it's mainly just DEV and RC environments, but um it's looking promising so far. These are some of the my favorite stories in terms of like customers who have gone in and expanded to a broader set of use cases. You've also spoken about our how our partnership has helped next even manage growth without adding to data center footprint and
29:16
like it headcount, can you share some more details? Yeah, so you know, today uh operates out of eight data centers across the United States. Um We have a total of 14 people on our infrastructure team that manages. Um And that's, that's not just storage, that's, that's uh the data center, actual data center management, that's network compute storage virtualization.
29:39
Um you know, everything that runs on that hardware. Um And that and that's no public cloud. So we're managing that all on our own. So, you know, peers really helped us from efficiency. It, it's just it's one less component to really have to, to manage. We also operate on the Cisco flash stack model.
29:59
So that's, that's been a real big benefit for us as well is that the reference architecture that Cisco and pure storage have put together really helped us ramp these data centers up quickly and manage them, um you know, really efficiently and easily. So, Eric our vision obviously as you heard in the morning as well is to have an all flash data center. Yeah, and part of that is the space and energy
30:26
savings that, you know, pr brings to the table. How important have those been for your your journey? Yeah, so and also as I mentioned, you know, we, we a lot of the voice infrastructure still requires bare metal. But um as we've been, you know, solely replacing a lot of our bare metal infrastructure with uh you know, converged infrastructure, the flash stack with Cisco and
30:47
pure, we've reduced our data center footprint probably down to 1/4 the size that it once was. So, you know, in, in, in many of these sites, you know, um multiple racks have just been condensed down to a single rack. Um and it's also allowed us though to go denser within the racks which generally becomes cheaper within the data center because you're only buying power,
31:10
you're not buying square footage. So, so uh I have, I mean, over this period of time was your capacity need also going down or you know, your capacity was increasing when you are actually reducing. Yes, our capacity is increasing. I mean, I could, I could state that in the eight data centers, we are actually not added any additional rack
31:29
space within the next five in the last five years, we've just reduced the footprint size of our infrastructure. So we've increased the amount of compute storage network capability that we have without adding any additional rack space and power has stayed, to be totally honest, has stayed pretty constant among those five years. So that's just amazing. Yeah, we hear about doing more with less.
31:55
We are. I mean, what once was a full rack full of storage is now, you know, three year worth of storage, which is, which is crazy and those are only half populated. So I could easily add another data pack and even add additional storage in the same footprint. It's amazing.
32:11
So, so you mentioned Evergreen in one of your earlier responses now with similar capabilities now available on our Evergreen portfolio across flash array and flash blade. I'll put you in a tough spot. Ok. What's your favorite part about the Evergreen subscription service? Yeah. So I think I, I probably wanna tell a little
32:34
story about um my first Evergreen upgrade and I think that and I don't know if there's any one in the room who's gone through an Evergreen upgrade. But um the very first time, it's, it's kind of one of those moments where you like, don't believe the vendor that it's gonna happen in that way. And I, you know, and, and they want to schedule during the day time and I'm like,
32:52
no, this is happening on a Saturday night at midnight, you know, and I'm informing the whole organization and we're letting all our large customers know there could be potential outage and expect expect the worst, right? And um I remember when, when I did the first Evergreen upgrade, I literally I met um at that time, it was our sales engineer who performed
33:10
it. I met him at the data center at midnight, we unboxed everything. This was upgrading from the um the 4 20 to an M. And at the time, pure had changed the chassis. So we're not even talking about just like replacing some components in the same chassis. We're talking about literally, this is gonna be a whole brand new storage platform.
33:27
So, you know, we racked it, we racked this new chassis and I'm like, this is gonna be unbelievable. I don't think we're gonna be able to do this. And so he, uh you know, he, he, he connects the cable between the two units, you know, we get the, the new storage controllers in the M at the time all up up to date. And he goes, ok, we're going to cut over one of
33:48
the storage controllers to one of the new storage controllers. And I said, ok, and I've got like a constant paying running and I'm writing files like nonstop to make sure that we don't miss any rights and, and he goes, ok, it's done and I'm like, wait, you didn't tell me we're gonna start, you know. And so, um, and I didn't even notice,
34:04
we didn't even notice anything happened and then, and then we fail the next controller over and not a single blip, you know, and then it gets to the point where, you know, with we had to move all the data. And so this is where I'm really thinking, OK, there's gonna be a problem, we're gonna be moving all of this data to a whole new data pack and you know,
34:22
he starts migrating drive or not, you know, module by module of data over and there wasn't a single, a single blip at all. Um And it, and it literally was like, that was the time, it was just, it was like magical and I'm, I'm not just exaggerating. It was actually like for me in being in it for 25 years,
34:40
it was like a magical moment to see something be as seamless like that and move from one storage platform to another. Um And then I'm happy to say that the next platform upgrade we did from the M to the X I scheduled it for like noon on a Friday. Like I didn't even really care, you know, when that was going to take place.
34:57
So it was um and you know, with the announcement today of the new uh revision four, you know, I need to talk to my, my team and I think it's time for the next evergreen upgrade, uh you know, coming soon um at some of our sites. So that's amazing. It's actually the second conversation I had today, which was similar someone came by the booth and said the same thing is like,
35:21
you know, you guys said evergreen upgrade the first time you said it, I planned it on a week and I thought you guys were just full of, uh, you know, just not reliable. So it was like the second time I did it, I was still like, I'll do it on a Friday the last few times I do. It is like, you know, I, I heard it as like Wednesday 9:09 a.m.
35:41
That's fine. Just go ahead do it. I don't really care. I don't want you to stay back later in the evening. So that's amazing. Uh So I wanted to get a little bit into APR one. So we hear from customers that pr one has, you know, really simplified storage management.
36:00
How is your experience? So for us, I think where pure one changed, the game was being able to manage all of our arrays in a single pain of glass. Um and visualize our workloads across all our data centers uh within, within one dashboard. The management has been a big key part. Also, we have other teams in the company or des
36:20
teams and some of our development teams who want to see how their workloads are performing. So been able to give them, you know, read only access to pier one to actually go in and actually see, you know, as they add additional elastic search notes, where are where are we seeing this? How are we seeing the storage perform? Um So that I just, I don't get that late night call saying,
36:38
hey, we did something wrong, you know, so they can look at it there. Um Additionally, we've, you know, really started using some of the uh the growth prediction capabilities of pier one to not even, not only see like growth as far as like how much storage we're going to be using potentially growing over two over a court period of time. But seeing how performance has impacted,
37:00
I think the performance component of it is is for us is a big part. Um to see, hey, you know, you're going to be hitting 90% utilization at this level, you know, is it time to upgrade to the next X model or do we need to do an evergreen upgrade sooner? That type of thing? So pure one's been it's been, I'd say simplify an already simple solution,
37:21
you know, storage solution to run. No, that's great to hear changing tracks a little bit. We at pr really are like talk a lot about it. Hear a lot about the impact that our direct flash module technology and the co innovation of hardware and software has meant for in terms of outcomes for our customers right now.
37:49
I don't expect our customers to be as in depth in terms of like designing the chips but being in the customer shoe. How has that kind of innovation meant in terms of experience for you. So I am not an expert in flash storage. I, you know, all I know is that at a certain point you're right enough to it,
38:10
it's gonna die, right? That's really all I know. So, um for us, I think taking the worry out of it is, is where, you know, and they, when they talked about it in the keynote today, um you know, about, about, you know, writing your own software, I actually have no, I, no idea what that means. Um But all I know is that peer has got it kind
38:29
of covered. Um You know, I do know that when we did decide to go open stack, you know, our devops team was really pushing for, you know, white box, commodity hardware, hyper converged infrastructure, you know, things that are like just buy stuff at best, buy and cobble together and give us nodes.
38:46
And I knew right away that was the wrong solution for, for what they were doing. Um And I was able to convince them to, you know, run our, our elastic and, and uh open stack nodes on pure and, you know, it's performed great and I don't have to worry about the flash behind it any longer. I don't have to worry about drives failing. I don't have to worry about any of that.
39:07
So, Eric one of the dimensions on which uh DF MS are actually uh much much better than hard drives or even other S SDS is a reliability. Yeah. How, I mean, have you seen any difference? I mean, I can tell you in eight years I've replaced one flash module, you know, 11 direct flash module in an array and that's it.
39:28
Um, maybe a couple of power supplies here and there, but one direct flash module. So I, all the, and I can tell you that the amount of hard drives and even just SSD drives that I've replaced in some of our bare metal equipment, you know, 100 200 times that. So, yeah, I mean, it, it to not have to worry about it.
39:46
Plus, you know, peer tells you when it goes bad, you don't have to figure it out on your own or through a monitoring tool. So, you know, Pier's got a module sent to you before you really even know that it failed or it's about to fail. So, no, that's, that's really great to hear uh from an actual customer. So, you know, there were quite a few announcements this morning.
40:09
I wanted to see if uh you know, what's your reaction to it just to recap. You know, we heard about the new um X and C. Yeah, refresh of that product line with the four. We heard about uh flash, a new platform getting announced and then 75 Terabyte, uh DF MS, you know, any, any reaction does it, what does it mean in your environment,
40:32
you know, we don't have tons of storage. So the E right now I don't think is a, is a fit for us if it was offered in a lower, a lower storage allotment. But the revision for controllers are going to be are going to be really big for us both on our C arrays and on our X arrays, we are reaching that performance level where we were looking at potentially going to the next
40:58
tier of would be we run X 50 today going to what's called 70. Now, with the R four S, we may not have to go to the 70. We can take advantage of it with the revision four. Also, we're on the first generation C arrays. So um we, we expect to see a major performance increase that and I think this is where like this is
41:20
where to tie in kind to the Evergreen is that for us, it's been beneficial because like we to think that I'd be running the same array in my data center for nine years. But even though the hardware has been refreshed, you know, 3 to 4 times over is amazing to not have to think about forklift.
41:40
So, you know, it kind of ties back to the evergreen. I mean, that's a great, I'm ready to go to the R four S and take advantage of Evergreen. Um You know, the 75 terabyte modules is huge, especially for us knowing that not, not that we may take advantage of that today, but knowing that we can keep that footprint size.
42:00
You know, as I mentioned, we don't have to continue to grow the data centers so we can easily swap out a data pack, go to the 75 terabyte and you know, and do that. Um And then, you know, I don't know, it wasn't announced at the, the keynote today, but I was in a session talking about the uh purity upgrades being able to be scheduled by the customer and the container basically the I
42:22
forget what they called it, but being able to now manage that on our own takes. Not that calling pure is ever a bad thing. Uh Yeah, the self service upgrades that's gonna be, that's gonna be really key for us and just even just another level of simplicity to manage um our raises and you know, and we can, we'll probably continue to add additional rays.
42:41
So the footprint is only gonna get larger and I think that's awesome. And that for, well, for my last question is our, our last question is around the upgrade. Sounds great. But overall, what does the future of the partnership between pr and look like? I think uh you know, additional arrays, um We're gonna continue to grow um the need for
43:08
additional arrays for additional data packs. Um We're on the original flash blade product today, so most likely looking at flash blade S is going to be key for us. Um, as well as uh port works. Port works is, has been big. We've been talking about it with our Debs teams,
43:25
um using it for database as a service for our Mongo DB instances. Um uh And then, you know, something I actually just learned about today and that is um uh metal is a service pure from Equinox product, which is something that I'm actually gonna start talking to them about now as well. So I think, you know, there, we've really in our data centers, we've replaced uh all of our storage
43:55
uh platforms with pure, but we still run a lot of bare metal. And so I think for us it's just, it's the migration of all of that bare metal furthering into more P A and I don't have any, I don't have any need to look at any other storage vendors right now because pure kind of solves the whole thing for us. That's amazing. And Eric, thank you so much for being here.
44:19
Before we take questions for Eric, I do have a couple of slides and let me try to do that without having a face plant. It would be unfair if I close the slide without plugging in a little bit about our new E family. So flash array E announced today and like, you know, really amazing product available at the economics of desk at like, you know, around 20 cents per gig,
44:42
including three years of service, irrespective of the protocols that you require. We have a product for you unified block and file for flash array and file an object for flash blade, massive sustainability, less E waste and much higher reliability and just a better experience than you get from your existing direct flash, sorry, spinning griff, not direct flash. Um If you want to learn more about e here
45:13
are a couple of sessions that you can check out tomorrow if you want to hear more from Manish and me, a couple of other slightly different topics that we are presenting in tomorrow are here as well.
