44:42 Webinar
Supercharge and Secure Your MDP Workloads with FlashBlade//S
Discover how to supercharge your backups with FlashBlade//S™ and our partners Commvault, Veeam, Cohesity, and Model9.
This webinar first aired on 14 June 2023
Click to View Transcript
Second intro will be good. Right. All right. There it is. Thumbs up and we can begin. Chad Monti. Is that our countdown? All right, we're good to go. Are you ready to talk about supercharging? MD P?
Jason, has anybody ever said? You know what? I don't want to supercharge or anything? I, I wanna down charge it. I actually don't want more. Never, never has ever happened. I mean, who wants to take the turbo off their car? I thought you put him on. That's it.
It is. I'm, I'm Jason Walker. I'm the global practice lead for modern data protection of data protection solutions near and wide and far all across the, the uh mythical RT orpo continuum. That's my responsibility. You get to cover everything across the globe. Now, I'm Chad Monteith.
I'm a principal. I cover it across the Americas, the field solutions architect together, Jason and I make data protection, supercharged, supercharged. That's right. So that's how we do it. What do you got here? This is you. Yeah.
Well, I was gonna, I was hoping you'd ask me a loaded question, but you know why, why pure storage for data protection? Well, I'm a, I'm a longtime backup guy. I'm very, a little too long. Frankly, I probably shouldn't admit to being in there, but, well, over 25 years, I'm a Florida man.
So math escapes me, but I'm pretty sure it's over 25 of doing, of doing backup. And I was very surprised about 4.5 years ago when pure storage asked me to come on as they were starting to do data protection because I knew they pure storage is all flash. You know, I'm used to as a traditional backup guy, slow and non-performing targets for backup,
right? It's just about optimized, you know, tapes and, and dis treating this like disc was a big deal to me 15 years ago on there. But the, the fact of the matter is pure was getting sucked in to conversations and why is that? Well, I mean, we could talk about some of the, the business reasons, reasons why uh of the support and,
and, but the pro the problem was that, you know, the, the, the legacy market was full of, you know, slower, single purpose, um you know, expensive to replace and update type solutions. But most importantly was the first one. This is kind of if you read in Hebrew, if you were going from right to left, uh the slower stores uh are out there.
And that's really the reason why I got some phone calls and not because I'm particularly fast, I think you probably look at me and figure, not a very fast guy but, uh, fast talking. But the, there was, there's real problems in the market and I've been in the business a long time. I remember, you know, I've been alive a long
time too and I can remember back in the, uh, the entertainment days when we would go to Blockbuster, you know, and, and we'd get, we'd have to go to a store to get a movie and then, you know, it wasn't even on digital, you know, we had the video tape and you'd have to, we'd watch it and then, you know, I'd have to rewind it or, and return it on time. Or else I get fined.
This was totally normal business practice. In 1997. The, the data protect the patent on the automatic Reiner, you'd have to slide it in. That was amazing how they did it so fast without breaking the tape.
But, but backup was facing the same kind of stuff back up and it was all of it and it happened all of a sudden, you know, like Netflix just kind of happened like overnight. It was, it was fast and the same thing was going on with data protection chad. I mean, we, we were going on, everything was optimized on the backup, everything was about getting it off host and just getting it to.
That's why it didn't matter what it was writing to because help me, you know, I never want to see and never have to restore. And if we do, we're doing restores. Chad. It was 12 kind of recoveries. It was much, everybody wanted to innovate and getting it off as fast as possible and
shrinking it down. And they all hoped and prayed that they would never have to touch it or restore it. I'm not sure many of them did. I, I, they were, the big concern was, can I get my email back? You know, not even the whole, not even the whole exchange, it was just getting a single email or a single file back.
Remember it was a big deal, you know, do you have the single mailbox restore feature? Oh my gosh, we need that one. So it didn't matter that that what we were writing to was really slow. But, you know, something amazing, amazingly bad happened, you know, almost a about 10 years ago, the, the specter of cyber criminals and ransomware came around and,
and it, it came around at the same time that we are restoring more data than ever. I think we all remember the recession 2008, 2009 people were asking me who's in the data business? Are you in trouble? And I said, are you doing more data or less data than you did a year ago? It's been the same. We're all doing more data.
We just saw, we saw it in the keynote earlier, more and more, more, more data. And so now the specter of having to restore all of it. And we were, look, we're looking at this pile of slow target that we'd been using. There was a problem and it was like, suddenly we had to think about chad,
we had to think about, you know, 50% 75% 100% of the data center needing to be restored. So, I mean, and then it, and even what we were writing to was, you know, there was these G farms. So it was taking up a ton of power and now we're facing with a, with a lot of stuff there. Then, you know, at the same time, it's like an origin story somewhere back in the hills uh
in parts unknown uh pure storage was dreaming up this uh target called a Flash Blade. It was, it's, it was a platform chad at tribute. Just a question for the crowd to see if you know, we, we've never talked about this before. I I might be in trouble. This might be the first time was Flash Blade built to solve the problems on the previous
screen. Was it built for me? The old backup guy. Was it built for that? It absolutely was not. It was not. No, they did not think of this brand new platform with all flash and, and, and screaming, they, they, they did not build it for data protection,
they built it for the next generation. Just what we've heard today A I machine learning, these, these processes that require super high parallelism and, and you know, super just fast, fast, fast. We were the technology version of a happy little mistake. That's right. Because, because one, at some 0.1 of our customers who had these say,
hey, uh you know, we got these sexy next gen flash blades in here for our machine. Can we also what we've noticed is that our databases which we used to be able to refresh in a reasonable amount of time now are are taking 36 hours to do it. And oh, by the way, we're in as a service company. So 36 hours is not tenable.
Can we, can we use these sexy next gen flash blades for for this? And of course, pure storage not known to be a data protection company that uh ok. Sure. And the rapid restore use case was born Chad. Do you know that that company went from 36 hour RT 0, 38 hour RT O.
The Davis is a half hour and did it going from what's on the left to what's on the right? So they're able to save space, save power, save energy, but supercharge their data protection. Amazing. Absolutely amazing there. And then what we found was that with the flash blade, uh especially classic flash blade, we were able to build out and the more, we did this, the better it better it performed the
scale was linear. So this is great for data protection. So that's when they called, that's what they called both the both of us. And um you know, and said, hey, we needed that because the, the, the, the specter at the same time that was going on ransomware was picking up and we had
to answer this question about restoring more data and less time. All flash was the answer rapid restore is, was, is and was the answer out there where the the rates can be up to three times faster than the rights? That is unheard of chad. Where's my restore tax?
No story tax. It's all gone. It's all, it's all gone there. And so, you know, it was, it's great and, and you know, so exciting. They created a brand new platform also not necessarily for data protection, but we're taking it. We're like the bad house casts.
Chad. You know, we moved in and like, you know, now they built an in law Suite Voice of Data Protection and we're hanging around because our customers have responded and they need this rapid restore. And as, as this has picked up Chad Shocker, I know, look at all of them, all of the partners wanted to come alongside of us because they don't have this a lot of,
they sell their own hardware, but they, they didn't have that rapid restore tier. They couldn't answer the question in a reasonable amount of time. So, you know, our partners came alongside and said, hey, can we, can we partner with you? Can, can we have a little piece of that rapid restore action? And we said, absolutely.
And we said, here's how you can make it even faster. It wasn't just throwing flash at it, but it was tying together the, uh, at the software level with many of these. How do we tweak them and tune them together? There's been a lot of work done over the last 56 years. Unbelievable. And a lot of it driven because our customers
found that new use case for us. It wasn't us just uh slinging hardware. I mean, I imagine there was probably a little bit of that but those customers saying, hey, we need to use it for this other problem that you haven't thought about and out of it was born supercharging MD P. Absolutely. Because the number one thing, as I know from a backup guy, the number one thing that I know about folks who are in charge
of data protection is that they, the number one thing they don't want to do is replace their backup software. No, it, it's painful. It's laborious. They would look, II I know that I would look for anyone when I was selling, when I was in an se for, for a couple of these back up companies and we were talking to
customers my number one competitor never carried a business card. They didn't even have a logo. Our biggest competitor was do nothing because it was agonizing to do it. So now we, we help that we, we're, we're enablers of a, of a do nothing culture and that you can do something but that something has put a flash
blade behind these data control planes. Boom. Doesn't every employee want their next year's goal to be restoring 1000 tapes and re backing them up. Yeah, you think getting that email back in 2005 was something. Now you see what they're doing here and, and doing it, doing it uh at scale. So it, you know, so popular that we said,
you know, the our customers told us, you know what we really want that you want more than just a rapid restore uh your use case here. And then so yeah, you know, we want the same thing but we want it at great uh even larger capacities in there. So, hey, look at that flash play at each ad being able to put come alongside, use them both at the same time, multi, multi dimensional, multi use case,
you can use it for data protection, but you could still do those uh those sexy next gen workloads. They're not not exclusive if you do them at the same time. So and you know, so now we have this, you know, slew of great outcomes that we can provide uh and, and at the same time, provide that rapid restore tier to make sure that being able to do
restores up to 70 80 90 100 terabytes an hour is relative for this day and age is that hockey stick of data growth continues. Uh What I love about this slide is not only are we running back up and restore on this, but you can start uh applying on here, your, your security event and incident management technologies. You can do your A I your machine learning.
Let's be honest, backup applications are not going to push a flash blade very hard. You have a lot of extra resources available on there and you should take advantage of it, reallocate those to something that is gonna extract other value to your business. That's absolutely right. And we'll talk a little bit more about some of the security aspects of this as well because as people started,
as companies started putting their data on the flash blade, they started asking us the question that you should be asking everybody that has anything in your data center. And that's how are you helping us in the fight the eternal struggle against ransomware? And you know, we have an answer for that. But one of the sad one of the things that goes
un un looked at when we talk about pure and data protection solutions is the evergreen business model and being able to take advantage of that and take advantage of all these bullet points here. The saddest thing is when I talk to customers and they listen and they hear and they think, yes, yes. Please give me rapid restore and then they realize they're stuck for two years because
they have some kind of contract that's keeping them, you know, locked into who they, who they currently have. But Evergreen gives you the ability to be more flexible. It's operational agility when you have this and, and this along with our, with pure one, our reporting over the top. Absolutely fantastic and,
and a hidden gem in any data protection solution. And without all the evils of uh the old E MC lease contracts that people used to write, I, I have a lease with the option to die. That was I think uh the, the thing that goes, I have 100 year lease on my 1999 storage. It's good. You know, I talk about all that and go through
and just kind of how we got here and why in the world pure in the data protection business, we've learned, you know, some things over over the years, some kind of key tenets. Yeah, we, well, we, we, we've definitely learned that's part of being uh being around is learning a lot. So we learned that flash blade was better than uh just an A I tool for MD P as well.
So when we think about what do we, what do we do with, with, with flash blade. We, we've come up and we've built these concepts here that we call the five ss of modern data protection. And really what they are, are properties that regardless of the data protection vendor you're using, we can use to supercharge any of those that, that list that you had up before con
Vault Coity V Spectrum Protect model nine. All the other ones out there that we've tested with, we can uniformly supercharge them with these five principles here. So I would advocate that when you're looking at your next generation data protection platform, what the next solution is, these five principles are properties that you should be
looking for that solution to have. And the great thing is you can add it to an existing solution or you can even rip and replace if you're on, you know, vendor A and you want to go to vendor c Well, make sure that that movement includes these properties and these properties are speed, security, sustainability, simplicity and scale. And I want to cover why each of these is
important. But we've been talking the whole time here about supercharging. So, so speed is almost self evident. We're looking at accelerating the typical restoration time by between four and 500% 4 to 5 X. From a property store perspective. I've got about a half a dozen slides that I'm going to show after this one to show some real
world examples with customers of what that looks like real numbers, real numbers, we like real world numbers or do we like marketing numbers? Which one do we want? Real numbers? Hopefully everybody's sitting there going, I just don't want to say real numbers, but it really is real numbers.
The second aspect that we need to inherently have is the property of security. We need to build that security pyramid foundation of ensuring that the operating system, that the configuration that the API S are inherently secure as sort of that first layer of protection, that first layer of immutability. But we want to add on top of that in case of emergency brake glass type of protection
mechanisms. The worst case scenario that you hope you never run into, but you're glad it's there if you do ever need it. And that's technologies that we have like safe mode, re instantiate your entire environment in 30 minutes. I mean, imagine what that was like. You know, back when Jason started working in data protection and Moses were in the data
center when the only vendor was pounding out. Back then, if you, if you think of his first uh restoration, it was probably, you know, get out and chisel a new server and you know, buy a spark five or something like that, you probably probably took a month to get there and then he installed solar. And what did you do?
He patched it, then you installed that backup. That was another few days. Then you found your MS D people and now we stood it up and it's like three weeks later and then you started restoring and everybody was happy because you met three weeks. But, you know, but Amazon was offline because uh they didn't have a pack of time, whatever, it was so bad for the business. We're bringing that down to 30 minutes.
Sustainability is the other key piece. You know, it's, it's amazing that the pressures especially in our European contingents are to reduce energy consumption. And you've seen a lot of the numbers, you've probably seen a lot thrown up on all the slides that we that we've presented over the last couple of days.
We have a number of examples where we've substantially reduced infrastructure, port cost as well as power costs. We had one bank that deployed three petabytes of infrastructure. What was amazing on that is that when by moving to pure, they dropped, it was about two years ago. So it was the the middle of COVID,
they dropped to their network port count from 700 ports down to 100 and 20 ports while five xing the throughput that they got on their backup and restore environment. And so I always challenge people. I was like when you especially back then have to go by. What's the difference of buying an extra 610 gig ports in your network? How long does that take you to procure.
And to be honest, even the biggest banks that are like top 10 accounts with the Ciscos and the of the world will still come back and say that's going to take me 15 months to get like you can't pull any strings to get that any faster. Well, that this can be the difference between doing a project this year or saying, oh, well, we're gonna put that on somebody's goal next year and maybe we'll get to it at
some point all of that. So not only are saving infrastructure costs, but oftentimes we're seeing power reduction in the 50% range depending on how it's done. We've seen 45 we've seen 60 it's in that wide range. So if you're looking at really reducing floor space, power cooling, um we, we've got examples and spades of doing that and in doing so you
get the scale and the simplicity as well. What's great about our data protection partners is how did this used to work before? Well, you used to install a new server, add some storage and you're like, wow, I need to, I need to map some Luns. That's awesome. Let's map some Luns to my backup environment
and then go provision and rebalance everything you don't have to, none of that exists anymore. Now, when you're adding with the object in the file protocols that we've integrated with all these, you add capacity and you slide it into your flash blade or into your flash or ac these applications are automatically detecting and using it the amount of work that you have to do
to expand these and scale. These is minimal. While getting the scale expansion, Jason was talking about earlier. There you go from one chassis to five chassis. It's not a fixed performance, you get five X the performance scale you can decide do I need to, what is my SL A doing here?
If my sl A is I need more capacity because I need to go from storing one month to seven months. I had blades. If my sl A is, hey, you're backing up at eight hours. But now you need to, are you restoring it to eight hours? But you need to restore it in one hour.
We can add infrastructure costs and increase that throughput simply by scaling it out. Every backup job, every restore job, every protection job gets the benefit of every component of that. When you do that, you scale it out, everybody is successful. And then because of that, we want to make sure it's simple.
All of the pure pieces, a tremendous amount of work has been done into the API S example. Scripts go look at github, there's more there than you wanna look at. You might just look at and say I wanna hire professional services. But if you're inclined, you can go ahead and absolutely do it yourself and see the example. So let's look at some real world data or semi real world data.
I think the first one, last thing I wanna talk about scale is uh we get, again, we're super charging, right? A lot of times the all these Softwares that we had, how did I have? Uh I don't wanna go too far back. Um All those, all those partners you saw on the screen, they all have some amazing function in their software. The only problem is the targets that they were
using weren't performing enough to make them reason any reasonable, especially at scale. No, not at all. Not at all. So pure was able to supercharge and unlock a lot of capabilities and all these capabilities they're putting in putting in chad, all these software companies are putting in are around security.
Every single one of the things they're putting in requires fast read. It requires that a tier of reading to make it happen. Otherwise, what, what point? Hey, if I've got it all, if I've got data and a panic room tucked away, but it takes me three weeks to get it all coming back out. What good was it to begin with?
Or if it takes three weeks just to read through it to look for a hash for a security event, right? How relevant is that? At that point? It might not matter. You might have already had everything on your production environment encrypted. So all of those pieces, all of that speed is enabling all of these successor things just
like Flash Blade originally came on board. And we found out, hey, it's great for data protection even if it wasn't built for that. Now, they're finding that pure under the covers of data protection is enabling a whole new slew of use cases that hadn't been considered before. That's right. Absolutely right. Now, more, more real numbers.
So these numbers were done with our Coity partnership here. These are what we call hero numbers. They were published in our lab. We've published a couple of white papers on our website that highlighted the environment that we use to reach this. What's interesting is all of these numbers. These numbers are big here,
4 to 5 times faster, rapid restore a petabyte per day of rapid restore. What this test profile demonstrated to us is that we can reach these numbers with actually very modest configurations. In fact, on the flash recover side, these, these were done with only four, I'm sorry, with only eight and 12 node clusters. Those are some of our more mid size clusters.
There's a number of customers that we have that are in the twenties and thirties and forties size clusters from a flash recover perspective on on the restore piece. So lots of performance there. This was an example that we actually did uh with one of our banks, we turned it into one of those hero numbers because you're now backing your infrastructure
by flash, it, it opens up those additional use cases and one of those additional use cases is usable instant restore. Now, all of our data protection partners are happy to say, hey, I'm going to sell you hyper converged. It's a hyper disk array and yes, you can go launch 1000 virtual machines on this. But what does that really look like from a physical perspective?
Well, you end up getting those virtual machines bound to a few spinning drives and I'm pretty sure that all of us with S sds on our laptops if we go uh launch our VM ware uh instance, and we spin up 100 V MS and we're sharing the 300 I ops that you get between these three drives that you're probably not gonna be terribly happy about it and terribly happy with the experience. So what do we do?
We change that? So that now all the IO is against flash blade and now your 1000 virtual machines have the power of flash behind them. Now, it's not perfect. Nothing's ever perfect. Right. We can, we can pretend it's perfect.
It's presentation. You are perfect. You are still going through, you know, engines and other components that are on the network latencies induced. So it's not gonna be as fast as if you were running on a flash array, but it doesn't necessarily need to be if you can get it built, if you can get it launched. If you can get instantiated that shrinks your L
the L A of how long does it take you as the storage or the backup infrastructure engineer to get these virtual machines up and running? Hand it off to your upstream application engineers where it's at least usable enough for them to start launching and restoring the application environment while we're doing the top piece there, which is the automated rapid restore storage remotion back to the original
cluster. All of this meant to make it usable so you can get your application environment stood up faster. We we have a customer uh a who was here on another segment. So if you're listening to this on demand, you can go and listen to the how uh pure storage has taken over modern data protection. And uh our customer uh Sean tells a great story.
In fact, uh after they've been attacked for a second time and they have all of this in place, they now have um this instant recovery process. They, they do not just 50 V MS, they do. I think it was 100 and 50 V MS every quarter as a part of their, their new process to make sure they're prepared in case of attack because attacks are not like the chicken pox chad.
You don't get them once and you never get again. In fact, it's the opposite. If you're attacked, chances are there's the, the blueprints out on you and you could get attacked again. So it's important to develop processes like that. We'll talk more about it as we talk about the security aspect uh as we go along.
And that, that after the example is, is a great one. You know what, what, what Sean talks about there is the first time they got attacked, they started the restoration. They were not using pure, they were not using rapid restore. They were not, they got like seven or 10 days into it.
And I don't think, I think they had like 10% of their environment restored, maybe less. It was so bad. They threw in the towel and stopped it and decided it would be faster to rebuild everything from scratch and to just start rebuilding virtual machines, that's untenable. He's like, how can I ask?
My employees took three weeks before they were back up, three weeks of downtime, which is the average, by the way for folks that aren't prepared uh in, in that way. Yeah. So he brings flash recovery. Uh or yeah, it was uh he brings restoring it whichever product it was. Um And he does that test that. Jason was talking about it once a quarter and
he does it all in three hours. So he goes from failing after two weeks to testing in three hours and then he got hit again because everybody who gets hit, they, all those Attackers are saying, wow, I've got a hot one. I'm gonna go sell what I did out on the market and make some more money. So I'm hoping to get a ransom and a finder's fee or something.
That's absolutely right. And they were so good. Listen to it for the folks here in the room that, that stuff is gonna be on demand. Check that out. You can pass forward past my nonsense, which is very similar to this nonsense. Uh But listen to our customer tell a real story there, real stories,
real numbers, speaking of real real numbers. So this was an odd motive. Uh customer that we worked with this year that was running uh and they had hyper converged storage, uh SD storage associated with it. And so we um we talked to them about the five SS. Hey, let's add speed, security, simplicity scale sustainability to it.
Show you what it looks like. They brought the infrastructure in for a test uh at scale. They were primarily using VM ware sequel file based workloads when they brought it in. They stood up that environment, the exact same infrastructure that they used in the current piece they were getting um in the current deployment, they were getting a little over 13 hours in order to do restore of their
environment. The only thing they changed was they wrote a copy of that data set to flash blade, kept the same infrastructure, same servers, same data paths, same targets everything else. Exactly the same. And lo and behold what happened.
We saw a six X improvement, a four X five X improvement. They went from 30 hours to two hours on that restore rate. The only change being adding flash recover into the middle of that they loved. It exceeded all expectations around there. So yeah, 85% reduction five times faster. The second example that we have here.
Yeah. Um was, was a spectrum protect. You wouldn't expect to see an IBM product in here uh around modern data protection. But it's true. It is there, there's a lot of customers that have spectrum protect. They don't want to get rid of it, but they do want to do what super charge charge it.
And in this case, we supercharged it with our block based technologies but that in in this example, supercharging it making it faster. Wasn't the most interesting thing. At least from my perspective, what I thought was most interesting was that from this was this was with a G SI who has spectrum protect deployments across hundreds if and maybe even thousands of customers. But when we built this with them,
we've actually published blueprints with IBM and and other components as well that highlight this. They did a safe mode, rapid restore recovery of the entire environment. So this wasn't just this was past the backup and the rapid restore piece. This was how do we re instantiate our entire spectrum protect environment from a safe mode protected snapshot.
What was interesting is. They said they have never got it to work with any other storage vendor. And they demonstrated across a whole bunch of variables, database lost directory containers lost, both lost. They're out of sync. All these different things.
The only storage system from immutable copy perspective that was ever able to bring back online that spectrum direct. In instance, in the data in a consistent manner was pure flash array. That would have been if you didn't say pure, quite not anybody else. It was just, that's uh I I thought that was the the most interesting out of that.
This is even more interesting to me, Chad and you, you played such a large role in this. I like this one. This is a fascinating one. So who would have thought that mainframe still needs to be backed up? It does mainframe is probably usually the most important data in environment and it's not really shrinking most in need of a super charge. Everybody is the most in need of a supercharge.
Everybody's like man, our mainframe just kind of runs over there and uh you know, and it just kind of just kind of goes and it's, you know, $60 million a year or whatever people pay for it. I mean, some obnoxious number, but we saw a lot of customers with mainframe, um who had the same problem. We need to protect the the problem that comes
to us as always, we need to protect mainframe data from ran some more attack. I'm not really sure a mainframes ever actually been attacked if it has, they're like, probably keeping it really secret and it was probably the IRS or somebody, I don't know. But, um, so we started working with a company called Model Nine who has a very innovative product.
It wasn't VT it wasn't based, they were on a very lightweight agent on the mainframe. Uh and they go over IP to back up directly at object store in this case, flash play. So this customer was actually running it to a hyper converged backup appliance. And those were the first numbers, the hybrid on premise, hyper converged S3.
Again, all we did everything the same we swap in flash blade. What did those numbers come out to the full backup and almost 81% reduction in run time? These look lower, it is lower. We actually hit the bottle neck on the network before anything else. And when you see a lot of these numbers, what's interesting is our bottleneck is not ever shown to be the flash blade.
Does anybody want to guess where the bottleneck is? It's always on the server so that the bottlenecks that we run into are on the target array that we're restoring to many of our tests. Like when we test these scale out on pure, we're not writing into one array. Many cases, we're writing like 8 to 10 of them to get the right throughput that we need.
But in this case, it was a network limitation, the archive run 97% reduction, the rapid restore, 92.5% reduction in run time to complete that. Yes, sir. Sure. So um very different uh protocol methods. So model nine is an agent that replaces like uh DFHSM and HSM and other products on the mainframe rights over the OSA
card to the flash blade directly. So those three components, software replacement, less software costs OSA card IP networking to the flash blade with Lumix valid for a lot of use cases where you have HSM or storage model methods where applications or the databases are directly addressing tape because it presents a native tape interface. So instead of an agent that picks data up,
you leave all your HSM products, you leave all your IBM software in place and it just targets a VTL end point on the icon which is presented by Lumix. Lumix picks it up, writes it over NFS to the flash blade on the back end. So where you choose between those two is dependent upon how other non backup applications are integrating with tape and how much you want to replace.
Do you want to get rid of your VTL environment, your and move to uh uh to IP or not? So really the big piece is how are you utilizing it? And are you utilizing in a way that one or the other can or cannot do no good question. So, because it's replacing software, how they, their big argument, their big value prop that they take to market is MS reduction
because you've got less software to license less MP charges. They actually offload the majority of what would normally be MS charges with like HSM and others onto the zip engines. Yes. No. And they seem to have been able to demonstrate some pretty, you know, some good numbers in that. I think,
you know, if you were to carve out the myths that you're, that you're using in there, um They've got double digit reductions so it's not in inconsequential, but it's good tech. So next one so that I don't run out of time. This one was another automotive customer actually where they were looking at replacing an existing backup environment for a GP FS system.
And when you look at the hyper converge the traditional way of doing things, what we like to hear 14,000 watts to run it, 1000 switch ports, 500 rack units in order to in order to operate that environment swapping out completely for flash blade. Look at this reduction here, switch port count, 94% reduction in switch ports to provide a greater amount of throughput.
A 65% reduction in rack units, 78% reduction in the in the wattage required to operate it. Um For this, for this particular automotive customer, substantial reduction here showing that all those things that we talk about, about reducing power, reducing cooling, reducing footprint, reducing auxiliary costs,
infrastructure costs, associated costs, networking, real data that we get from customers. When you're looking at doing a supercharge, this is a great way to, to, to think about it too. We have customers that have thought, well, you know, maybe we should change. I'd love to supercharge what we have. I don't know if we can afford it.
And then you take a look at what you're what you're saving on floor space and cooling and all the things that are important because energy audits are a real thing and you have your h back audits and everything else that, that you're doing, you have numbers, you have to hit. This is what we talk about when we're talking about sustainable and then,
you know, now you have a predictable platform, not just for this year but moving on as well. Yeah, and that's how we start to reduce that 2%. Now we talk about all that and we talk about, you know, you have to wrap all that around a security story. There's nothing in your, in your environment that you don't ask the questions how it's going
to impact in the case of, of a a ransomware attack. And we like to, we like to help folks talk about things before the attack. You want something simple to implement, you need something that's immutable uh and resilient and, and the thing that I've seen change the most, I'm taking full credit for it. Chad you and me,
babe. Absolutely. Is that more customers talk about restore now than they did four years ago. And again, for the reasons that we talked about. So you don't wanna some, some solutions will offer one of them, some solutions may offer it two of them, but you need them all. We're, we're gonna, we're gonna put that on the record on your Wikipedia page.
That's it. And recovery is the one where a lot of solutions fall down, fall down on. And here's uh here's something you want to read later. But these are all the things you you think about and before during and after the attack, some things don't have anything to do with pure, but we like to help folks prepare and again, just like our,
our customer at Apne, it's about getting that cycle this whole cycle tighter. Every time you go through an event make it tighter. The chaos that ensues. Uh you know, during the attack. It, it, you know, that was one of their first attack, Sean shared because I,
I asked him, I was like, how was, how much was the chaos? He said it took a couple of days just for, to get someone to make a decision to do something and Sean had to step in. It wasn't even his position. He said we got power all this off. They they weren't even prepared at that level at that point because you don't really know until you,
you know, Chad walked over and punched me in the face. I don't know how I'd react. I don't wanna know by the way, I don't, I don't wanna, I don't wanna try that out. What stays in what happens in Vegas stays in Vegas, right? But, but you really, you really have to think about, about all these things and when we talk
about af after the attack, we talk about definitely wanna focus on recovery and we definitely want to talk about all that stuff that as far as cleansing and clean room, all that stuff requires fast read to be able to take, take care of that. Then you can start thinking about things like putting your logs and getting snapshots of that and, and proof and, and what is your legal team running?
Because that's a key part of that whole cycle because they're gonna have to do review and they're gonna have to do a lot of that. And I don't know if you know this, but legal teams aren't up to date on what's fast, they'll use what they have and it can be terribly slow and, you know, not to not uh no defamation of a, of an entire community, but they're not all that concerned with being fast uh on what
they're doing to get the job done. But if you can have their, their stuff running on all flash because again, it's not a single purpose, then you can get that done faster. Now, that cycle is tighter and tighter. Uh Every time, every time we go around security is, yeah. So we like to talk about the uh the security frame framework that we've got across MD P here.
So across the board, when you're evaluating data protection partners, when you're talking about uh data software infrastructures, there are these five pieces here, these five properties in here that you want them to possess, that you wanna have in there. You want to make sure that the platform you chose that you've chosen has sufficient guard
rails built into it. Does it have immutability built into its file system? Is it deploying on top of a secure operating system? Does it use a secure version of Linux? Have they locked it down? Have they followed best practices on locking down Linux and operating system instances? Are they utilizing encryption across the board?
Do you have safe mode integration or if it's right into object storage? Do they have integration with technologies like object lock where you can force that immutability onto the storage system? And does the storage system have an immutability implementation where people can't override it, where it's a compliance enforcement versus the more traditional enterprise enforcement model?
And then you go up and, and, and, and look at how the access control is. Your typical AAA services is the API locked down. Where is it open? How do we generate keys? Uh Do they have back doors into the system? Can I apply appropriate access controls, an integration with active director and other components that I need to tie into?
Do I have compliance certifications? Do we have s certified encryption algorithms in there? Uh If we're going into federal instances, do we have authorization to operate those pieces that are key to validate that we've got a, a secure system? And then as we move up, do we have the ability now that we've got something that's fast to add
additional value on that? Can we do anomaly detection? Do we have the ability to, to use algorithms or machine learning components or you know what we now call machine learning A I to determine clean recommendations? And once we determine a clean snapshot, can we do a rapid restore? Can we even bring it back and make it useful to customers and have them actually have a
restored environment from that demonstrably certified if you will clean version of the backup? So wrapping things up, we've talked about the problem. We've talked about how what the solution is and what you you should be thinking about here. Remember it, you know, speed, security, sustainability and simplicity all at scale.
You know, all adds up to the sixth which is supercharging. You remember supercharging and doing it with the operational agility of pier one chad mentioned anomaly detection. I just heard Matt Bradford from our pier one team talk about anomaly detection and being able to detect DRR reduction and everything else in there. It's all hands on deck.
Generally in your data center, it's all hands on deck on peer, all of us are working to make it more secure, make you ready for when that happens and get you back faster than anybody in the industry all across the resiliency tiers, snapshots protected by safe mode as sync active cluster active dr all of that protected in safe mode pure as everything is pure as a source protected and pure as a target
as we've talked about here with flash blade and, and all of it all. Yeah, all of it adds up to a uh a five S solution and I was gonna add that we're finding uh that customers are finding it valuable and important to establish those process and those run books uh early on because there are so many locations that you can find and investigate this data pure we're exposing, hey, here's clean snapshots on your array.
You wanna do the fastest restore possible. Just a snapshot restore maybe. So could be. All of our data protection partners are running analytics on their platforms and feeding it into their cloud. Do you want to go look at your backup partner and see where it is and do an instant restore. A rapid restore? Maybe.
But I would encourage everybody to think about it beforehand. What is the order that I want to approach these in? Where do I look first? Where do I look? Second? And what's my prioritization? I'm probably going to prioritize pier one and a snapshot recovery over going to a data
protection partner and having to go back 14 hours instead of 30 minutes things to think about beforehand. And why is it important to think about beforehand? Because if you're in the middle of an attack, the last thing anybody is probably going to be doing is thinking straight about what the order is. So practice it, have it written down, know what
you want to do. Yeah, that's it. So all across that continuum ending with flash blade, your last line of defense, your backup repository for rapid restore and security with, with safe mode. Take advantage of that. If you have any questions, Chad and I will hang around afterwards,
be happy to answer any of them. We're weirdly uh interested in talking about these things. So please please feel free to come up and you can pick on Jason and figure out what he was doing 25 years ago too. You don't wanna know, just ask him for a history. That's it. So for Chad Monteith, I'm Jason.
Walk everybody on listening to us on demand. You missed a packed crowd. They were wild, they were rowdy. We had to have people do crowd control here. But uh it was a good, good crowd. We appreciate everybody being here. Thank you. Thank you.
We will build on this with our security pyramid and share how object lock helps organisations apply rapid restore to a secure and immutable backup copy. We’ll wrap up with how FlashBlade//S can be used for forensic analysis, helping to identify the known gold copy of the data to rapidly restore from.
Test Drive FlashBlade
No hardware, no setup, no cost—no problem. Experience a self-service instance of Pure1® to manage FlashBlade, the industry's most advanced solution delivering native scale-out file and object storage.
Meet with an Expert
Let’s talk. Book a 1:1 meeting with one of our experts to discuss your specific needs.
Questions, Comments?
Have a question or comment about Pure products or certifications? We’re here to help.
Schedule a Demo
Schedule a live demo and see for yourself how Pure can help transform your data into powerful outcomes.
Call Sales: +44 8002088116
Media: pr@purestorage.com
Pure Storage, Inc.
2555 Augustine Dr.
Santa Clara, CA 95054
800-379-7873 (general info)