00:00
Welcome everyone to our session and accelerate. Um My colleague and I my colleague, Ray and Paxton and myself, Trevor Starns are gonna be presenting on powering modern scale out applications with the ultimate object storage solution. So uh again, my name's Trevor Starns. I'm part of the Flash Blade product team for
00:25
our global field strategy and growth. And uh Raymond is a technical marketing manager in the on the flash blade product team. So when you look at the industry, there's main categories of how you access storage. Um And you're probably familiar with most of these because you're here at a storage conference but block storage where pr got our
00:50
start fiber channel Izzy super simple, low latency, high performance workloads best suited for structured databases uh VM ware, but it relies on the host or the application to to handle the file system in the in the metadata and all those things. And then obviously you have file storage which have been around for decades,
01:13
things like NF and SMB uh leveraging TCP/IP protocol, high throughput requirements, large unstructured data sets where you don't have an application actually structuring that data. Uh but it is higher hierarchal um object storage is a little bit newer in the in the market. Um also leverages TCP/IP protocol but more of a
01:39
standard H GDP DP and rest API for access and for workloads with massive scale requirements, high throughput. But it's flat, it's a flat data organization. So you don't have some of the complexities at scale that you see with with file storage. So the growth in object, excuse me, object storage is significant specifically
02:06
for all flash. We'll talk about pure stance on this, but there's a lot of things driving this. Um And you see here, there's a lot of statistics around organizations that either already have all flash for object or they're looking at it. Um a very high percentage of the market and a lot of those are because applications are driving that interest.
02:26
Um And for those of who have deployed it, they've seen a significant advantage whether that's simplicity or performance or scale. Um And they call it a game changer. So we'll get into some specific examples later and I mentioned some of these. But if you look at why uh what some of the benefits of object storage are we mentioned the flat structure and that's really, really important at scale.
02:51
Um And unlike a file system buckets provide that and don't have as many limitations when it gets to scale, simplified metadata, addressing objects through user and system metadata performance, some of that performance is because of that flat structure. Um And you know, even if you look at it compared to traditional backup applications,
03:15
which a lot of them are now leveraging object storage. Um you know, we're not only seeing on flash, consistent right performance, but also very consistent read performance that we didn't see when we look at the legacy object platforms. We talked about scale and we talked about simplified access protocols in the early days of object if you go back.
03:37
And I don't know if anyone in the audience remembers this, but there were some proprietary products, proprietary API S. Um And you know, that's uh you know, with the emergence of AWS and other cloud providers, GCP and Microsoft Azure, um more of a industry standard way of accessing that data and we're seeing that move into the on prem world.
04:01
So I mentioned earlier applications driving some of this. And if you look at some of the, you know, obviously there's been a lot of cloud native applications that have been developed in the last decade or so, but a lot of other traditional applications are making that move to, you know, handle the large highly complicated data sets that they have.
04:22
And if you look at products like com vault, they've moved to object as their preferred repository method. VM is another one that's new, that is more recent. And you also look at even analytics applications like Splunk, Splunk went from a what they called Splunk classic to smart store,
04:40
which went from block storage and NFS to object buckets. And you can do a lot of really much more flexible things. And, you know, also it's dis aggregating compute and storage, which is a trend in that analytic space which allows the, you know, the compute nodes to scale much easier.
04:59
Things like maintenance of a compute node much easier because it's decoupled from storage and object storage helps, helps provide that capability. And then you have other categories like custom software development. Uh We have a uh a customer who is a provider for the recruiting and staffing industry and they just had too many files to manage and it was getting very
05:24
complicated, very difficult for backup and dr very difficult for management. And they, they recently moved to object for their platforms, they rep platform that and they expect to get significant more scale out of that because they are expecting growth. So that's a very common theme when you look at some of the fortune 100 customers providers, CD N, et cetera.
05:46
And then even databases, you see some examples here post Maria DB oracle, which can now leverage object. And I, you know, I would, I would say those are, you know, more emerging and, and even Microsoft sequel now supports S3 as part of their backup. Um If you're doing native sequel backups,
06:08
you can actually point those to an S3 repository. So where does pure fit into this? Well, we're going to talk about the architecture here in a minute, but we made some big bets and it's really been recognized by organizations like Gartner and Gig, where Gartner sees us as, you know, not only a visionary in this space but also high ability to execute in that
06:28
distributed file systems and object platforms. GIG looks at the balance among features and performance and scalability, also some notoriety from those folks. So what is it that we're doing? And what why did we see this as such a big opportunity back in the 2016 time frame when we launched Flash Blade? Well, you know, if you look at traditional legacy object platforms,
06:52
they're very difficult to manage at scale, right? Even moving to object, if you look at most of them, uh the legacy platforms are based on disc designs, um some of them even leverage object interfaces on top of a file system which doesn't really solve the ultimate problem it solves the how do I connect to a problem but not the scale problem or the performance problem.
07:14
And you know, in most cases, they weren't designed for high performance, they were designed for content repositories, archive tiers, those kind of things and you know, they weren't really using storage resources or capacity efficiently, right, that really plays into, you know, in general, why we've been so successful and why we're so excited is around,
07:35
you know, some of the things that we're doing with direct flash modules and the evergreen business model and getting maximum efficiency out of the hardware that we use. So why does performance matter? Um we saw this coming and um you know, if you look at things like uh in, in the analytics space specifically in things like security logs, right? If you're using Splunk for security logs or
08:00
elastic or something like that, the ability to retrieve or get search results faster is incredibly important, right? Um And from a compliance perspective, PC I socks, and you see a lot of the other verticals where regulation is is really important to be able to provide this in a certain amount of time, you know, and what does it do to the brand of the organization if you can't retrieve
08:21
or provide that insight in a fast enough manner? So this is another really big theme around what's driving a performance need in all flash. So we talked about these but operational efficiencies, we really think that storage not just object, but all storage should be incredibly simple to manage, incredibly simple to maintain. It should deliver high avail availability
08:43
throughout hardware and software upgrades. And we think that, you know, we talked about inconsistent search performance as well as well. Storage should scale seamlessly and be able to provide the support in a very consistent manner. So some examples are, you know, if you look at the at a software development workflow, this is a really good example J frog artifact uses files and objects.
09:11
And this particular customer upgraded a multi chassis flash blade and replicated over almost half a petabyte of data in a weekend, sorry, almost half 400 bytes of data in a weekend. In the automotive industry, pretty much every manufacturer is looking at things like self driving autonomous vehicles. And you know, there's a tremendous amount of data and analytics that goes with that.
09:35
We have a customer in that industry and they are migrating hundreds of applications across multiple platforms with object that makes that much more simple and possible in semiconductor. There's a story similar to that SAS provider I mentioned where the size of the objects are too large and they're too complicated with NFS and that's caused performance issues and scale issues, right?
10:00
Um In some cases, these organizations aren't able to put that data in AWS and they need an on prem, an on prem platform that's capable that's fast, et cetera. So I'm gonna hand it over to Raymond for a deeper dive in use cases. Thank you, Trevor. Um So we're gonna go into some use cases that
10:19
we're seeing today that are utilizing objects and then some emerging use cases. So quite commonly, one of the first uh use cases that was seen for object was actually data protection. The ability to actually back up data, an object was really used as a cheap and deep storage. Um With the advent of flash blade we actually brought object so that we can actually provide
10:45
rapid restore with object. So now you get that high performance object built in. And most backup providers today either allow for object to be used for tearing off from your primary backup source or as a primary backup source. So this gives the ability to actually back up to object for rapid restore capabilities or with our E platform.
11:09
Now being able to actually have long term archives within there. But data protection isn't just about backing up the data to uh source and be able to recover it. It also you need to have the ability to actually protect that data itself. So we do have ransomware protection built into the platform for object as well as object lock capabilities in the platform.
11:30
So you can protect that data that's actually being backed up so that it can't be deleted in the event of a ransomware attack. The next area was modern analytics. So previously, analytics was using S3 as more of a archive backup location or off loading data that it didn't need within the hot tier anymore.
11:54
It was just sending it off somewhere cheap object storage. Well, with the advent of flashlight, they're able to store very large quantities of data and actually be able to analyze that data directly from S3. So now they have that high performance ability to analyze large quantities of data that traditionally with file. You would have a lot of complexity,
12:16
to program, to a lot of complexity, to store that data very, very easy to actually store and get real time analytics out of object and especially an object with flash. Another one we see is a lot of custom software development. So a lot of times go and talk to software developers, they're developing something to AWS. You ask them,
12:35
what storage platform are you, are you actually using on AWS? So which service? Oh, well, we want to use object. OK. So you're using object. How's that performing? How's that working? They go. Oh, it works well enough.
12:46
But we want something fast. We want it on premise into our own data center. So we can bring that development uh project that they're working on back to the company and be able to actually put it into production. And this is something that we offer with flash blade is to be able to provide that near tier one performance with object. So those software developers can take that ease
13:09
of utilizing the API S where you just put get list, there you go. You're all done. You're able to actually access all your information, right and pull out your information. Very, very common use case out there today. And with the introduction of E we're seeing a lot more of these data repositories.
13:30
So long term data retention, long term data repository on S3 and especially with a large scale amount of data out there today. So a few references out there. So JO CD N ended up using our object storage to be able to store large amounts of cash CD N data and not just store that large amounts of cash data. One of their issues was being able to delete
13:58
that cash very quickly. And so they are actually able to tie directly into the flash plate API and with fast delete, able to delete that data extremely quickly right out of the system itself. So CDN being used to actually cash all the information but also needs to be wiped and refreshed constantly. And they're able to actually take advantage of
14:20
our S3 that way as well. Seeing de op shops being able to actually improve their work flows going from weeks down to days or hours to be able to improve their workflow capability out there. And of course backup environments um being able to utilize R S3 either for rapid restore or with E now long term retention.
14:45
Some of the emerging areas that we're starting to see out there is a IML now, it's an interesting area because it has a lot, a lot of data. And so being able to actually send that data into S3 makes it very easy, very quick. They can have large scales of data without any problems that you would normally have on a file system within an S3 bucket itself.
15:08
Um And this makes it so that they can store all that data access, all that data and be able to run their models. The only one disadvantage that we're seeing with this is S3 doesn't perform necessarily at the exact same performance as NFS. You got all that overhead from S3 itself. However, when it comes to simplicity of being able to store that data versus trying to create an program that will actually store on a file
15:34
system, all the file system background that you need to create all the different file structures, access all the file structures, uh figure out how to traverse that as an application. This just makes it a lot easier where you may have some overhead. But with flash blades S3 capability, that overhead doesn't necessarily play that big of a role.
15:56
So they're now starting to see a lot of people programming towards R S3 to be able to actually have A I data stored within S3 itself. Another interesting one is enterprise imaging. So uh enterprise imaging is kind of interesting because depending on which application you're using, sometimes it's a lot of really small files, sometimes it's very big files that you end up having stored in a very complex
16:21
directory structure that's created by the software itself. So most enterprise imaging software today is looking at how they're going to leverage and take advantage of S3 because it can simplify how they've built those destruct directory structures and how they can store that data and re access that data. Also, they can take advantage of a lot of the capabilities of S3 such as adventure
16:45
consistency where it will automatically replicate that data from one bucket to another bucket in another location and still be able to access it. And with eventual consistency, it doesn't mean it's an instantaneous replication. It will actually replicate that data as soon as possible. So they don't necessarily need that imagery data at the other site instantaneous
17:09
because that doctor is not going to look at it within a minute or two minutes. They're gonna look at it a few hours or a few days later. So now you have that data located over there and they don't have to build that into the application itself. So they can take a lot of advantage over what S3 offers for them.
17:24
Another very interesting place is database and database is an interesting one because normally everyone thinks of a database as how low latency can I get access for my database? Because that will improve the performance of my database, not always the case with every database, but when you look at database with S3, there's a lot of things like a time series database where now with versioning within S3, you can actually have a time series database
17:51
where every time you put that new record into the to the actual bucket itself, it now creates new version. And so now you can just go back and look and it's automatically built into the protocol itself. So you now don't need to actually worry about programming any of that stuff, it's built into the protocol.
18:10
So it now gives you that ability also to for more of a traditional database, you now actually have version history, you have other information that you can get that's built into S3 itself. So this one is starting to become more and more interesting project that people are looking at. How can they take advantage of some of the advantages of S3 and put databases into S3 itself.
18:34
So we'll probably start seeing more of this coming out over the next few years of different types of databases. A lot of databases are already programming to be able to back up to S3 such as MS SQL. Um And then there's a few databases that now offer drivers in the back end to be able to actually utilize S3 as the primary storage protocol.
18:56
And there's also brand new database types that are being built that fully leverage S3 out there. So a lot of interesting stuff in that space. So with that, I'll pass it back to Trevor. Thank you, Raymond. So we talked a little, we hinted a little bit about what we're doing and obviously performance and all flash is a big theme of
19:17
that. Um But when we built Flash blade in 20 we launched it in 2016. And leading up to that, when we built the product, we made some really big bets those bets, one of which was the industry was going to need fast object, right? They needed a platform that could leverage both
19:35
file and object on the same system because in many cases, moving to object was an evolution over time for a given workload, right? They may start with something like NFS and then over time they need to, they need something that's S3 compatible and they don't want to have to go buy a whole new storage platform for that.
19:52
And that's one of the beautiful things about flash blade. They need something that's really easy to manage and can support a large amount of unstructured data. And you look at I'll get into the architecture here in a minute. But if you look at purity, it's a distributed key value store can support. I think it's something like 64 quadrillion
20:11
objects or something like something you would never reach, right? Theoretical. It's actually a limitation of the hardware right now. It's not anything to do with what the operating system can handle and it can do that at scale and for both file and object. And and obviously, the third thing is performance is critical and specifically not
20:30
just topline performance but performance in a very mixed dynamic um uh diverse set of workloads, right? You might have low latency, high and another application that needs high throughput, uh et cetera, et cetera. So when you look at the key features and capabilities of flash blade that we've developed over the course of from G A till
20:54
today, you know, Raymond talked about object safe mode and object lock. That's something that we've developed in partnership in some cases with some of our early customers who are thinking this way about being able to recover from a ware attack with immutable, not just immutable, slap snapshots but snapshots that or points in time or in this case, objects are simply locked and to prevent a rogue admin from being able to do anything
21:23
with that data, right? Inability to delete that data based on a policy compliant with S3. We talked about that and so out of the box application compatibility with AWS flexible security models. So I'm not going to go super deep into that, but you know, security of the object interface and the product itself is a high priority for pure and
21:44
we talked about scaling. So I'll skip over that. I don't want to be too redundant. And we did talk a little bit about object replication. We can actually not just replicate from flash blade to flash blade or even flash blade, S to flash blade. E first gen flash blade to E first gen flash blade to s doesn't matter.
22:01
We can also replicate to AWS natively. So that's a really good feature for anyone who wants to be able to move data into AWS. Maybe it's, you know, a temporary software development effort or something that's being tested or something like that, we to steal a little bit of Raymond's Thunder two Fridays ago in our most latest dot release 4.1 0.5 we also now by special
22:29
approval only will allow customers to replicate from Flash Blade to other third party compatible S3 platforms on Prem. So you think about some of our competitors maybe have some other reason to do that. And then, and then the other thing which is also related to replication is multi site ridable buckets. This is really mostly applicable to spunk smart
22:49
store where you have a multi site configuration. In this case, we support up to two sites today and that's essentially an active, active single name space. Uh As far as spunk is concerned when you're moving, that is a feature of spunk smart store. Um And again, that fits into that, that, that use case that we talked about earlier about
23:09
things like security logs and, and uh you know, one of the leaders in that space. So we'll be here a while if I read all these. So I'm not gonna do that. But a significant amount of R and D and engineering from a software perspective has gone into our object capabilities and you can see the list of them here. And we launched object about, I don't know if it was two or three quarters after the protocol
23:35
launch, it originally launched with NFS V three, NAS 4.1 S and B as well as object. But object was, was the second one almost immediately after NFS. It is native object. It is not on top of a file system and I'll show you a visual of that in a minute. But if you look at these releases, all of these releases were possible with no planned down
23:56
time, they're part of an evergreen subscription. There is no software licensing. We don't have software licensing on flash or flash blade. So um and you know, not just S3 features you see here, but you also see at the top, we've significantly increased the scale of the platform where today we can go up to and Raymond will get into some of the details of
24:18
scale actually in a future slide. So we'll hold that one and so a little bit deeper dive on the S S3 A P compatibility is things like object level operations, bucket level operations list, multi part uploads, virtual hosted style object versioning, all compatible on flash blade. But we also have some features that are unique to flash blade.
24:44
Object lock itself is no longer a proprietary thing, but we do some enhancements around freezing objects and safe mode for bucket protection for things like, you know, one of the probably most common examples on flashlight is combat and being able to object lock on backups for ransomware mitigation, that extra layer of protection. We're also doing that now with VIM version 12, which is pretty new and exciting um massive
25:10
object sizes we can support. We've tested up to 90 terabytes, 50,000 parts, which in AWS compare that to five terabytes and 10,000 parts, right? And fast delete Raymond mentioned this earlier. We've actually had multiple customers that said Flash Blade was the only product that could solve this for them where they do massive deletes in big large batches and we can do that
25:33
incredibly fast, which gives them time back whether that's time general operational time saved or potentially in a lot of cases, especially in software development type use cases where it's actually helping them get to market faster because they can build and develop new releases faster than their competition. So overall, here's a look at the the logical
25:57
architecture of flash blade. And I mentioned that it was designed around this massively distributed transactional database, right? It is not a traditional legacy file system. So it's actually built on a key value store that scales significantly larger than the hardware can even support today. And we do provide NFS and SB file system capabilities as well as three object on the
26:20
same platform. No need to go buy an entire new storage system. No need to have a separate shelf of flash just to do both. It's something that customers tend to really love. Whether that's because they've consolidated multiple use cases on the flash blade or if they've evolved over time and the applications which in many cases we've really hammered that
26:42
home have been moving to object in a lot of cases, distributed everything, metadata, objects, dynamically, self tunes and self heels. There is no need to put load balancers in front of this product product, right? As few as one virtual IP address all the load balancing and distribution happens on the back end automatically.
27:01
There's nothing for the administrator to configure incredibly simple and obviously, data protection built in. So you can lose multiple blades, multiple dms and continue running. And we also do support non disruptive upgrades between generations. That's a new thing that we're actually going to go in September.
27:19
But we, we'll be able to go from first gen flash blade to flash blade s non disruptively. And in the future, we'll be able to do that as well. If you buy 200 blades and you want to go to S 500 we can do that non disruptively as well without having to replace all the land and repurchasing everything. So it's a really nice capability and program.
27:41
We talked about diverse workloads. So variable block size, small or large objects. Again, no load balancers and flash blade S and E first gen didn't do this, but flash blade S and E now leveraged the same architecture for the direct flash modules as our flash array platform.
27:58
And you know, from a company perspective, that's giving us a huge advantage in terms of driving density improvements over time. You've heard a ton about the 75 terabytes and we're going to just keep going. Um And from a data services perspective, it does go well beyond just object, but we've talked about a lot of these and you know, one thing I haven't mentioned is things
28:23
like compression and data reduction, which is something that we're continually trying to improve the efficiency of that over time. For example, when we launched S 200 we launched a more aggressive compression algorithm which improved data reduction on average across our fleet in a significant meaningful way. Um We're also looking at doing some things in the future to improve and drive that down
28:43
further. So this is just a quick graphic on going from first GEN flash blade to S 500 I believe S 500 we can obviously do well, maybe it's not obvious on first GEN flash blade, we had a blade with processor complex memory and man, there was no separation of those. It was, it was a blade that was the field replaceable unit.
29:07
When we moved to flash blade S about a year, we decouple the NAN from the compute and the memory. So it allows us to do multiple configurations of ratios of NAN to compute on the platform based on capacity and performance requirements. But it also allows us to get better performance. We not only increase the CP and the ram of the blades, but we increase the networking and the amount of DMS,
29:32
we can put into a blade which allows us to get significantly better performance in flash. And so Raymond's gonna go a little bit deeper into the market. Yeah. So within the flash blade object market itself, there's actually a very diverse market, a lot of different industries that are right now, either utilizing S3 or planning to be utilizing S3 very soon.
29:59
So things like the banking and financial sectors, telecom gaming, retail, health care, and are all industries that are either looking at or are using three today and are planning to actually use it for a lot of different use cases. So some of the ones we've already talked about modern data protection with VAT and V uh modern analytics, uh small smart or, and others.
30:25
And then uh when we were talking about the custom software development, that's where this whole des application modernization. A lot of developers want to use a fast object, they want to actually get in and use it. And if you talk to a lot of industries, they're looking at how can we leverage it and developers want to leverage it and use it and uh of course,
30:46
infrastructure as well. Some of the infrastructure stuff we're seeing out there are becoming use cases where infrastructure organizations where they develop their own C DNS or their own applications on the infrastructure side are starting to use S3. 1 that seen very recently is taking all of the log data and storing all of that log data that's coming in from multiple locations and all storing that on to S3 buckets and then
31:12
analyzing all that data from it. So they're ingesting it all into S3 and analyzing it off S3. Um So where did we typically see object? Well, normally in the past object was sitting kind of where this dis area was uh when we introduced flash blade that introduced fast object. So we had fast object out there to provide a
31:35
lot of new use cases. Well, with the introduction of Flash Blade E, we're now able to actually also compete in those large scale repository uh workloads where object used to be used very heavily for archives and for long term log repositories and whatnot. Now, the E platform is built off of controller chassis and expansion chassis.
32:01
Today, we support up to four chassis within the system and we just plan to grow this system significantly over time. So the controller chassis act as both processing and storage. So that's what receives all the requests and the expansion chassis just start adding on. So you can add storage to the system and keep growing the system by adding storage itself.
32:26
So we do offer it in three flavors, small, medium and large. And of course, this will continue to get larger and larger. So extra large, extra, extra large and so forth and so on. As we continue to add more chassis to this system. So today raw, we can do about four petabytes on our small end.
32:43
And then we get to the large end upwards of around eight petabytes in capacity raw within the system. And then we have compression that goes into it. Now, the E platform is ultra dense, so very, very dense platform and very, very low power. So it provides a great environmental efficiency out there today. So you can get up down to one kilowatt per
33:09
terabyte with the E platform and significantly denser than most competitive products out there. And I would add a couple of things just really quick is first of all, we've talked a lot about 75 terabyte DF MS here at the conference this week. Those will be, those will be G A at the same time as on Flasher A E here in I think a month or two, a few months. I can't remember but that will increase our
33:33
density even further and we will continue to increase that density. The other thing that Raymond covered was the concept of control, chassis and expansion. And over time as we get more dense and larger, it's not something that our customers should have to worry about configuring. It will be linear price per terabyte. So you don't have to worry about OK,
33:52
I hit some threshold and all of a sudden I have to spend a whole bunch more money for compute nodes. That's all part of the pricing model. To be linear as we grow and you get all the goodness as well. With this product of the everlasting software upgrades, all the new features that we're putting out for flashlight are going to be in this product as well.
34:13
So just kind of hidden on the use cases here. Not all of these are object use cases, but we can see where the S fits versus the E some of them are such as data protection. The S is meant for that rapid restore capabilities. Those business critical applications you need to get back quickly, you can store that on the S platform and then the long term retention on an E platform.
34:36
So you have now that option for S rapid restore E for long term retention. Um And then all the backup vendors do support generally either NFS as well as S3, either for long term attention or as the primary analytics. Analytics is one where you generally are going to store any hot data on an S platform that you want to do a lot of analytics on a lot of queries or reports.
35:04
But the long term retention generally was tiered off to a dis based system. Now, with the E system that could be tiered off for long term storage on an E system out there. Instead, um Another one to, to hit on here is the A I and ML. So people that want really super super high performing A I uh building models with deep learning. Um That right there, you generally want to
35:31
actually store that on an S system. Usually NFS, those that want the easier, easier access would store that on S3 itself so that they have can program to, it can get access to it. And then you can actually store the rest of the data after you finish learning your model on it into an E system and you can always bring it back to the S if you need to,
35:51
to retrain it. Um A few areas we haven't seen a lot of S3 on is HPC and ED A. Those are areas we just haven't seen that much S3 more so NFS on those areas now comparing it to AWS. Um So today, if you're utilizing AWS standard for S3,
36:14
that's where you can end up using flashlights and get tier one type performance on S3 on premise and just change from instead of targeting AWS for your S3 access, just target the flashlight. And then at that point, you can actually just move your application that you started developing in the cloud, move it on premise and have that tier one performance for production.
36:37
Um If you're using it for more along the lines of something like glacier or infrequent access, that's where E fits fits a lot better to be able to store that data. Keep that long term retention of that data as a repository at the end of the day. So just kind of a chart to see where things fit if you are currently using AWS today, if you're using Glacier Deep Archive, well, that generally would go to tape.
37:02
That's really the cheapest of the cheap um that you have within there. So those wouldn't really fit within the E system unless you're looking to bring it back on premise in house. Now, a few things with what's new with flash blade. So a few of the key items here. So object replication.
37:23
So we do have three site fan in object replication that recently came in. So now you can actually fan in from two flash blades back into one flash blade. So that gives you the ability to fan in your data from two locations back into one. Um You also can replicate from any flashlight system, whether it's first gen or S into an E system. So it gives the ability for you actually to
37:48
replicate your data. So you have data out at a remote site that you're collecting the data, you're analyzing the data, then you could replicate it back into an E system to keep as long term afterwards. And um as Trevor mentioned, the ability to actually replicate to a third party S3 target with approval. So you do need approval to be able to replicate
38:14
into a third. And specifically, we we've been able to support replication to AWS for a long time, that's not new, that does not require approval. This is for on prem third party other non pure storage object lock is another one that's fairly new that we did implement recently. So it does offer similar capabilities of A S plus those enhancements that we added to it on the flash blade.
38:40
Um and then object quotas. So account to bucket based quotas, hard and soft quotas. If you want to go more into details with the within any of these new items that are out for object, I would recommend talking to your account team and they can go into detail on to each and every one of those items and how they function, how they work.
39:02
And then lastly, if you want to see similar sessions on flash play, there's one coming up in Jasmine B um that you can go see right after this.
