00:00
All right, everybody. Thank you for coming today. Um Let me um we're gonna talk a little bit about pure storage for critical financial sector needs. I know it's a little bit different um than the title on out there. Just bear with me. We're gonna walk through a couple of use cases that we've had success in as a pure partner.
00:20
And we really wanna show you what we can do with pure and how we partner to build better solutions for you. So I'm gonna talk a little bit about Kendall. So if you're not familiar with, with Kendall, we were formerly ibmgts, right? We are now a separate independent company. We're going through a revolution of changes
00:45
internally and we want to remind you that we're here and we can do uh anything that you need in strategy adoption and optimization, wherever you're at in your solution. We are a services company. We are no longer a hardware or software company with services. We are purely a services company.
01:08
So we have approximately 90,000 staff of which uh the vast majority are technical people who are ready to help you with any of your solution. Uh design analysis, whatever it may be, if you've got an existing solution and you need us to optimize it. We can do that too. Um We have 30 years of designing, building and
01:41
managing mission critical it environments under our belt, we have 35,000 people certified for public cloud, right? And it's not IBM cloud, it's Azure, it's AWS, it's GCP. We have skills and experience in all the public clouds.
02:05
Uh We have 22,000 people certified for Microsoft products and solutions. We've got 9000 on Aws. That one is still coming up and we've got 4000 on Google also still coming up. We have been around a really long time in one form or another.
02:30
Uh So we are, I like to say we're the world's biggest startup, right? So we were 19 billion uh when we spun out from IBM and we're still on track for that. We still have five of the top 10 airlines. We have five out of five for the top automotive companies by
02:58
revenue. We have 11 out of 20 of the top insurance companies. We have four out of five of the I can't read this of the top consumer uh retailers. So we have a lot of experience, um our recent earnings announcement. So we've got about 75% of the fortune 100 right? Give or take a few percent.
03:26
It varies almost every day to give you an idea of the scale that we have grown. Uh Over the last two years, we have 576 exabytes of backups that we take annually. We have increased our footprint that we manage storage uh into the
03:54
thousands of exabyte range. So storage always grows, backups, always grow and it's going to grow more. We have a coming revolution in A I. That's about five years out. But you've seen announcements from companies like a MD where they have processors focused on
04:17
A I, right? No more GP us repurposed as A I engines. It'll be purpose built product. OK? So that's coming. And when that comes all the data that they have, all the instrumentation that exists for A I is going to need a place to be stored.
04:42
You want that storage to be energy conscious, you want it to be environmentally friendly, you want it to not occupy the whole data center as it once did. So now we're at a place where we can get storage down into a couple of view of space. So I mentioned that Kendall is going through a lot of changes internally from the way that we used to be. So we're now agile, we're digitally
05:18
enabled, we're business minded, we're self sufficient and we're continuously improving. The business minded is the one that changed. And my, the biggest change in my book because we consider your business as a services company equally as you do, if not more. Because your success means our success. We're no longer a product company. We're here to provide you a service.
05:45
We have changed how we operate entirely and still we're still doing it. Um We were just having a conversation earlier about how we're changing how we approach client engagement. So you'll see more changes from us, but we are not the same company that we were. You may have seen press about Kendall Bridge and how it unifies your experience.
06:15
We have direct customer access to bridge coming then a matter of weeks to months, right? It's coming. So within the last two years, we have developed this entire ecosystem. It's designed to make it easy to integrate. There's no more of the traditional lift out your tools and put ours in. We'll take what tools you have in place.
06:39
If you need tools, we can provide those, we have recommendations, we have standards, we have tested them all. Uh We know what works, what doesn't, but this is about easy integration and easy operations. So bridge is the path into your environment, right? It's a isolated control plane and data plane. So your data never touches the
07:05
control plane. It is completely isolated and it's only loaded if you have access to view that data and that's the case internally and externally. So if your Coke and you know Pepsi is there too, you cannot look at their data. And if you support Coke, you can't look at Pepsi's data, right. It's all isolated.
07:29
Uh It's only loaded on demand. There's no big bucket of data for someone to log into and look at you have to have access and you have to request it explicitly. Uh through bridge, we support uh people processes and best practices. We have a best practices, compliance engine. Um It's always evolving.
07:55
Uh This engine has well over 1500 rules currently. Um It's usually oscillating between about 1518 100 active rules that are checked daily. This is across multiple vendors, multiple technologies we look for things like password links or how your uh storage is configured. If it violates any best practice there,
08:21
make sure encryptions on, make sure that you've taken a backup in the last 24 hours, right? We flag all of that stuff. It's all displayed via bridge and this is the path into uh the environment for you and us. So the the client view will be on very, very shortly. Uh But it gives us a unified console,
08:43
it gives us an API and it gives us a catalog of services and as soon as it launches shortly thereafter, you'll actually be able to go in. And if you need something that's not in the catalog or in the base service, you can request it, right. You can request it and we'll work with you on, on developing it and it will be deployed in short order, right? It's not going to be a 6 to 12 month product
09:10
cycle. It's going to be weeks. So we also have our partner integrations included at bridge. So uh by year end, we'll have full integration for pure um edit data uh including performance and capacity data that will be directly in bridge.
09:34
And we'll be able to um provide analysis on that data um trending, right? And you'll be able to dig into that data, download it, do further exploration on it. So let's talk about one of the wins that we want to cover. So this is a financial services company.
10:01
Um They had selected flash array X along with us to replace their legacy storage environment. This was planned to migrate October to January. I'm sorry, it ran from October to January. It was actually planned until March. This was two months ahead of schedule. Why?
10:26
Well, we ran the project pure, brought the tools and the staff. So equipment was stood up like that within less than a week. Um This company is heavily regulated. They have lots of restrictions on when we can and can't do stuff and, and alter things in the environment.
10:50
So we had very limited time to do this. The fact that it finished two months ahead of schedule is a testament to how easy it was to install the pure hardware, get it up and running and then get the data migrated, right. So we coordinated all the Change Windows, we coordinated all the other service areas but putting in the flash array, XX was easy.
11:15
Like the, if you've looked at one of the quick start guides for your flash array, it's like four steps. It's crazy. Um So in total, we finished 63 days ahead of schedule, right? And it wasn't because we had people working around the clock, they obeyed all the change windows, they obeyed all the change
11:37
restrictions. Um So, you know, one or two hours every week was what they were allowed to do. It was, it was really restrictive. Um The legacy system was able to provide roughly 2 to 1 on their data reduction. The peer system provided 4 to 1 with their data. So you'll see a lot of claims and guarantees about um data
12:04
reduction rates from various companies. OK. So pure always stands behind theirs. Uh We have standard set internally at Kendall to know what you're gonna get with particular data set. So if you're wondering like does my Sap Honda get 4 to 1? No, it it doesn't, it's 2.09 to 1.
12:25
Um So, but Hannah is already pre compressed. So not all workload is this particular workload is a mix of uh V MS and databases. Um So databases are typically pre compressed, but we were still able to get a 4 to 1 overall out of it. It was great.
12:45
Um So that was doubling the effectiveness of the storage. This particular company had elected to actually refresh a little bit early because they needed to comply with new regulation that said their data had to be protected. It had to be recoverable. Uh It had to be recoverable in an insanely short amount of time.
13:10
Does uh got some coworkers up here. Does anybody know what it is off hand? Like what they were required to do? No. Ok. So it was, if I'm not mistaken, it had to be the entire environment within like four hours, then it was a pretty big environment. So um it was 824 physical and virtual servers, 1.6
13:37
petabytes. So do the math if you have to recover 1.6 petabytes. How long is that gonna take? Right? If it's not cyber protected with a snapshot, you're not getting it back from tape that quick, right? Um Encryption at rest was also a requirement. The company had not previously had
14:01
encryption at rest, even though it was a feature on the previous product they needed it. Um So let's talk about this and let's highlight the things that really worked well here. So as Kendall and as pure, we worked together to bring what we needed. So the pure, the Kendall team had knowledge of the environment.
14:28
They had the relationship with the customer, they knew all the ins and outs of the change management system. So they coordinated everything, they planned it, they planned the waves. Um And here pure brought the tools and technical staff to install and then trained the local team.
14:52
So the local team here, um we do a lot of enablement internally and we try to target teams that we know are going to be supporting product. Um We do general education sessions and we do targeted in this case, the team got a little extra, right, because pure brought in education for the team as well. So there was no bumps on switches of
15:21
technology. The team was just as comfortable with pure after the training as they were with the legacy session technology that they'd had for years. Um In this right, um This particular company has a requirement for availability that is pretty stiff. Uh Previously, they had a very complex uh set up that needed to
15:54
be managed to make this work with pure technology, it was able to do it with just the base product. So um in excess of four nines is what's required for this customer. Um pure is currently at seven. So I'm gonna stop here. Uh And uh I wanna take some
16:19
questions. Does anybody wanna talk more in depth about anything that we have talked about or this particular case speak? Just see. So port works is a great solution for containers. Um This particular environment is very heavy VM
16:52
and VM only. So I don't know that it's a requirement here, but port works would give you portability right of your data across a physical environment like this or a public cloud. So if you have a container environment where you've optimized your applications for a quick restart and recovery and everything comes up in a matter of seconds
17:20
port works would be a, a good solution for this environment as well. Um But in this case, it was a pretty heavy legacy environment based on a BM architecture. So we didn't change the architecture but we changed the technology supporting it. Um Why did we do that? Well, there wasn't enough time to comply with regulation. Um It needs to be reworked still and we may end
17:45
up at Port works here, but for now the customer has made an investment on pure that is paid off. Um They have the up time they need, they have all the feature functions they need and we were able to do it just over two months ahead of schedule, which is unheard of in this space. Anytime you deal with a regulated customer, you know how restrictive it is. It's just rough.
18:07
Um Sometimes when you go over change windows, you have to get regulators sign off, right. Uh I don't like talking to regulators. They're mean mean people, sorry if there's any regulators here. Um But a, you know, this was just a wonderful turn key solution that we brought in stood up in an unexpectedly short amount of time and we were
18:33
able to really deliver for the customer, the innovation they needed because they got two new capabilities that they had not had before. So I'm gonna talk a little bit about safe mode. So safe mode provides immutable snapshots. Um This is purest terminology for it, right? Um You can instantly snapshot any volume, zero reservations or planning required zero
19:03
performance overhead. Um You've got flexible consistency groups. There is a lot of good here. Um I would like to say that this is the best. Um But legal told me I couldn't.
19:21
Um So we'll see about that. Um Think about this, right? So if I have 1.6 petabytes of data across almost 1000 virtual and physical machines and I lose it. And a regulator tells me that I have to have it back in four hours or I'm out of business.
19:47
What do I do? Well, this is what you do. You use safe mode, you're not gonna take any performance hits on this, right? All you're doing is buying extra capacity. You don't buy a new license, you don't have to install new software,
20:04
it's already there. So all we did here was come up with a smart schedule that allowed us to meet the objectives set by our customer. And now we have protection, we had to size the arrays properly to start with, right. So you're gonna need a little bit more than 1.6 petabytes if you want to protect 1.6 petabytes, but that's it,
20:34
right? Um They're always thin, always de duped, always compressed. You get the same space savings that you have on regular volumes on your snapshots. They don't take up more than what you already are spending. So it's great.
20:53
So you can recover uh forward or backward on any volume. Uh We do quite sorry about the echo. We do quite a bit of um testing and verification before we bring in a technology partner. Um We have an entire suite of tests. Um
21:21
I believe there's 49 tests that we do on functional and operational things. We also do firewall um definitions, security reviews, we define data flows and how it has to behave. We create um procedure document for our delivery teams. We do global education sessions. Um We validate that the product is cost effective because um we're not gonna pass
21:53
through cost to you, right? It has to be cost effective. Um So we only put in stuff that we have thoroughly vetted. In this case, we did two separate under load upgrade tests because this customer had a history of problems with their prior solution. They had a, a history of problems with their prior solution because of how long it took
22:23
the pathing for that solution to recover before the Os recognized that it was recovered. So the Os 27 seconds was the time it allowed the time it would queue before it went down, the storage took 26 seconds. So on some occasions, it didn't make that 27 seconds and it went off line. So, fairly large oracle based cluster, I believe it was rack in particular.
22:55
So they had lots of issues with recovery and stability in the environment. So that was one of the driving factors here. When we looked at it, we said, well, we need cyber protection. OK. We got that covered. We need to make sure that um we have things that the admin
23:18
can't delete because sometimes we are worried about bad actors, right? Especially unregulated stuff. You don't know what somebody's gonna do and they're off hours. So you want that protection, you want it locked down. So we tested all of this stuff. We test every product that we have very,
23:38
very thoroughly. In this case, we did a full uh sap um workload while we did an upgrade, we took down nodes, we add a new code, we did everything to it. No difference. Uh We did that in concert with pure, in pure remote lab because we didn't have in ours.
24:05
We wanted to test Hana because it's one of the workloads that people have the most trouble with. It's a little bit esoteric to size. Um It's a pretty demanding product because workload comes in waves as the memory pressure mounts, shoves everything out to storage. We also did a try to break it upgrade test in our lab.
24:31
So we generated artificial workload and then we went and yanked out parts while it was live um, stuff you're not supposed to do system didn't falter. Um, there might have been maybe a 1% difference. I am. It was totally fine. Right. Every other product that we do that test with, they fail so
24:58
big win for that big win for being able to lock the snapshots. So the admin can't delete them that guarantees protection for the customer. This is super important in the financial space because you can't lose any of that data and you have to be able to get it back quick or you're not in business anymore. Next case I'm gonna talk about and this isn't the exact solution
25:30
here. What I'm doing is I'm showing you what you can do. This is really close. OK? So in this one, this is a bank. So instead of a financial services firm, we have a bank here that uh needed a new solution. The bank had problems with replication from
25:53
site to site. Pure implementation of H A is different than other vendors. So instead of the more traditional, let's take all our data and push it as fast as we can through the network to the other site which came out of mainframe. We have a very smart, very production focused
26:17
implementation here. Pure says, hey, production site, I need all your data over here. Can I get it? Production site will say, yeah, sure. And then sometimes it'll say, hey, my loads increasing. So I'm still gonna give you the data but your
26:36
RPO time is gonna increase, right? Does it increase gigantic amounts? No, it doesn't. Um, there's a, uh, 11 millisecond round trip time limit here, but you have a sliding RPO scale. So if you start out at five minutes, maybe one day you'll get to eight and then it'll go back
27:00
down. Is that ok? The knee jerk reaction from customers is usually, no, that's not. Ok. Well, I have worked lots of impacts um As far as I know the second most in the company. And let me tell you, no one cares about H A when production takes a performance hit,
27:22
no one, any time that production has a hiccup, no one cares if there's a second side or not, they just want production to perform because it's impacting their business live depending on the company that could be millions of dollars an hour, right? So you want dependable production performance while you do replication to protect your business, that's what this
27:52
bank got. So it's a relatively small solution. It was some fifties, four of them total. Ok. Uh But that was still able to accommodate all the banks data and workload. That's was unthinkable just a few years ago, right? Because of the effective capacity that they're getting out of this,
28:18
they're able to replace it with four total boxes, two of which are mirror targets. They had a super complex H A design before now because the storage is doing the heavy lifting. It's easy. It looks pretty much what I have up here on the side. Uh The difference is uh that they don't have a rack,
28:46
rack two here. Um So they have the primary site and then the stretch I put this up here because this is what we would recommend. And I think this is what pure recommends. Uh You do when you have this type of setup where you need a really quick recovery to another site in case that site goes dark.
29:12
Um VM Ware VD I is very well suited to pure. It does very well with it. We consistently have 4 to 1 or better reduction rates just like the previous client. Um There's no performance impact here to protect the environment. That's another problem that we have a lot is all these replication solutions all depend
29:37
on pretty hefty processing, right? So there's always some impact to production. Maybe you insert some latency into every IO usually 1 to 2 milliseconds or maybe you can't run as many ops or as much throughput on a single product as you would like. In this case, there is no difference whether this solution is on or off performance is the
30:07
same. Um Just like the previous client, there was a security need here. So they were able to um meet the regulatory requirements for security. In this case, it was encryption for data at rest um also in transit. But uh the way they had implemented, it was, was done at the uh switch level. So um the key
30:35
here is the third bullet down able to provide sub millisecond IO at four gigabytes, a 2nd 92% capacity usage for each X 50 deployed. So if you've dealt with flash for any amount of time, you know that there's some recommended overhead because of the wear leveling and the flash, ok. Without the wear leveling in the flash, you could go up to 100%.
31:01
But because you need to rearrange data occasionally so that you get even where in your memory then you have to reserve some 92% is pretty good. Usually the number is between 85 and 90. So 92% is, is really pretty good here. Um Since they had two pairs of X fifties, this customer is now able to push eight gigabytes a
31:25
second. They notice no difference in their production workload and they have their H A well within the tolerances, the regulator gave them and they can recover very, very quickly compared to what they used to. And now they don't have, I think it's four staff they had managing all the software that used to do this.
31:47
Now it's done by pure hardware. All right. So I'm gonna talk a little bit about um the Kendall private cloud infrastructure as a service delivery team. So this team I used to run from a technical standpoint. Anyway, Dave's got it. Now.
32:14
Uh This team, I'm still very, very proud to have worked with this team and to have helped define it. Um This is the team that we will have work with you on pure implementations. This team has the only NPS 100% that I'm aware of.
32:37
No detractors. OK? We have no detractors on this team and that team, the team has maintained that day for how many years? Five, OK. Five years. So over five years, no detractors uh for a services team,
32:59
I want you to show me somebody else who has that. Look at your cable company, they're gonna be like 68. If, if they're good, uh look at your car company, they're gonna be in the fifties. I mean, nobody has this. Uh The team is excellent. It is built on a strong one,
33:18
team culture. So the culture for this team is what we're trying to put into everything now. Um They are built and designed to be very supportive of each other. They are built and designed to pick up skills and then pass them along to the team. Um There is no, well, it's his fault or, you know,
33:43
it's will's fault or it's Jennifer's fault. It's, there's no finger pointing, right? The team owns it. So the team supports each and every member. Uh they have an exceptional uh track record with um availability. So this is end to end, OK? Regardless of the technology.
34:07
Um whether it's, it's pure or something else, whether it's, you know, brand A or brand B switches in the environment, whether it's ice or fiber channel, it doesn't matter. So this is an all inclusive number since 2021. Um They track it multi year because it's a statement. Ok. This is how good this team is.
34:33
They track it multi year every couple of years, they roll it up. We used to track it from 2018. OK. Uh But over five nines um across hundreds of environments and customers. So um I'm supposed to tell you that the environment's definition here, right? So when I say environments,
34:55
I mean, prod DEV Q A customer or data center, OK. So could be any of those. When I say environments, you know, the lawyer said I had to, had to say that uh because it wasn't clear, but this is a fantastic way to get your pure services delivered.
35:23
Um Charlie had up that slide this morning in the keynote about how cheap everything was. This team will provide you all the services and all the software and all the hardware for one price and it's actually potentially lower. I gotta figure out what unit that was this morning, but it's potentially lower uh than what Charlie even had up.
35:51
And um it's just uh an amazing team, the leadership is located in the US. Um And then the rest of the team is split across India, the Czech Republic, China and Brazil. Um I will say that there is some leadership in the Czech Republic as well. Um He's one, he's one of the leads that does uh defense work.
36:14
So he's awesome. Um The team has skills across the board, not just storage. Um Although they do specialize in it, you cannot do better for a service delivery the than this team, in my opinion.
