00:03
Welcome to our ask the experts technical session on two groundbreaking innovations from pure storage, real time, enterprise file and zero move tearing. I'm Joshua Petty, more commonly known as JP from product marketing. And in this session, we'll dig into the innovative features and benefits from our recent launch at pure storage.
00:25
We'll use an FA Q format with me asking the questions and our product managers and experts providing in depth answers. Throughout this video, we will explore how these technologies work their key benefits and how they can help your organization achieve greater efficiency and performance in data storage and management.
00:46
So with that, let me introduce our experts, Kartik, Srinivasan and Jonathan Carnes to tell us how real time enterprise file and zero move tearing can transform your storage infrastructure. Karthik, thanks for joining us. I know we want to get into the technical details. But can you give us a quick explanation of zero move tearing?
01:09
Thanks JP. Zero. Move tearing is an innovative approach to data management that eliminates the complexities of traditional cheering. It allows for efficient balancing of hot and cold data performance needs within the same storage array. Customers need a really smart way of tearing data. Without all the complexity of moving the data
01:30
across disparate storage systems or tiers. We are excited to announce an industry first concept. Zero move tearing pure storage is innovative approach to providing the TCO benefits of data tearing without the complications of the various moving parts associated with traditional storage systems.
01:49
With zero move. Cheering flash blade can provide access to hot and cold storage classes automatically within the same name space all without dealing with stubs, licensing multiple management planes or copying data to separate storage infrastructure like legacy solutions do. So that's really interesting. How does it work?
02:15
Well, let's take a look, zero mode. Turing works by intelligently managing data based on access patterns. It prioritizes hot data without physically moving the data around ensuring performance, predictability and cost efficiency. You know, think of the storage being disaggregated from the logical compute and
02:41
network resources. Once we identify that a particular data set is hot, the system serves it quickly. The cold data on the other hand does not get the system resources at all and is uh slowly my apologies. Um Z MD uh as we colloquially call it internally ensure data is accessible from the same name space using the same endpoint.
03:08
So there is no disruption to clients. So that's very impressive for a customer. What are the benefits of zero move tearing. So the key benefits include uh performance predictability, cost efficiency, infrastructure, agility, and simplified management. And let's look at these things. Uh uh you know,
03:31
one at a time now, performance predictability, as I mentioned uh that comes from uh dynamically allocating the resources to serve the high performance data needs. So you are not moving the data around, which means that you always have access to the data. And you can actually uh pick up the hard data at the highest performance that you need cost efficiency. Now, because we combine both uh you know,
03:56
very uh performant uh flash plate systems along with capacity expansion shelves, we are able to get cost efficiency and we're able to handle both hot and cold data performance needs. Uh Without any of the complexity and overhead of traditional theory, you're not managing more than one system. Uh you're not uh dealing with complex policies.
04:18
All of those things help. The third is infrastructure agility. You're able to seamlessly transition workloads to lower performing storage classes, prioritizing new mission critical workloads. And lastly, all of this leads to simplified management.
04:34
It eliminates the needs for complex data movement reduces operational costs. It's really, really a very simple system. So that sounds like it's a lot of benefits for customers. How is it different zero move tearing compared to traditional data tearing solutions? A great question. JB you know, traditional data clearing
04:57
solutions often involve multiple uh storage systems with different types of media. You know, you have uh all flash for high performance, you have disk based systems for low performance. And they put these things together to provide TCO benefits. They also involve moving the data between separate physical tiers which which can
05:20
introduce inconsistent performance, operational complexities and costs. When we spoke to a lot of customers, this was one of the biggest aha moments for us where they felt like, hey, they had to constantly readjust the policies to make sure that the system worked the way they are workloads needed. Zero vol tearing. On the other hand, manages data within the same
05:40
array, avoiding the need for separate control planes, reducing the overhead associated with managing multiple systems. It also does not move the data. And so uh you do not have the uh you know any of the performance penalties associated with moving the data back and forth.
05:57
Uh And so, like I said earlier, uh uh you know, it is within the same name space with the same endpoint. So clients don't see the difference at all. Zero. Move tearing also eliminates a lot of complicated node based uh data movement licenses, making the procurement and deployment and management. Uh an integral purity FB feature.
06:19
That's excellent. So what capabilities will we offer at launch with zero move tearing and what can our customers expect later? Sure. Now zero move tearing uh is going to be launched in waves and in the first wave, we're gonna launch what's called pinning of file systems. Now, what really happens here is that file
06:45
systems can be manually pinned uh to performance or hot uh in a storage class or to archive or cold storage class. Customers can also transition the file systems between the performance and archive storage classes with a simple uh click of a button note that at launch, we will only support whole file systems object support, which means support for pinning
07:08
buckets will be in a future release. And in that release, you can pin entire buckets to a performance or archive storage class. Now, that's not all we will actually continue to innovate uh and offer additional capabilities in subsequent releases. Now, if you take a look at the slide, this is what we will be working on.
07:29
After we launch the first wave in a future release, we will do what's called granular uh uh you know, file system or object support. What that really means is that get differentiated performance at an individual file or object level. Customers can set a policy at a file system level uh or a bucket level indicating that any file or object uh that is older than a time period uh is old and cold and needs to be
07:57
treated as such. The system will then determine the appropriate performance for different individual files or objects based on this particular policy. It will, it will take a look at the metadata associated with it the access time, modified time to determine whether a particular asset a file or an object uh is really uh needs the highest performance or can be treated as cold data and the system will
08:23
appropriately uh uh adjust the performance. Now, what will happen also is if you access the cold data over time, it will start becoming warmer and it will become harder, the more you access cold data, it just gets hot and it'll get you the, the highest performance.
08:38
And the reason for that is when we first access the data, we don't really know whether you're just doing it one time or whether you're actually doing it more than one time. So the subsequent accesses will make it warmer and hotter and that's what we plan to offer. Now, I will tell you this all purity features such as snapshots,
08:56
application, safe mode, uh warm immutability data threats, encryption will apply at launch and subsequently as well. So since we're launching pinning first, uh like my biggest question is how does that work? Right. Let me go back to the, the previous slide and I'll walk you through. What, what that really means?
09:19
How does spinning really work? And that's what we're gonna launch at G A. Well, when you create a file system, it is automatically set up for high performance, all the data is considered hard until the cus uh unless the customer explicitly tells the system that it needs to be called, now, they can do that by simply uh you know, using the GU I or the rest API or the cli uh you know,
09:40
it's a simple operation. You actually attach a policy saying it is cold, it's a one click operation and, and once that is done, uh it will immediately transition to code. Uh you know, you can also transition it back from cold to hot, but that will be slower uh depending on the size of the file system.
09:58
Now, once you do your pinning, whether it's hot or cold or you know, in, in other words, performance or archive, depending on the storage class, all the data in that file system is treated either hot with its performance, which means that it gets the highest performance or cold, which means that it gets very low performance. Now you're allowed,
10:17
uh you know, all you can pin the file system and the capacity for it will be determined by how much performance capacity you purchase JP OD. Awesome. Thanks. That's a great explanation. So when we look at other flash blade products, how is zero move tearing different from those flashlight products, flash brade systems with zero move tearing are multi chassis,
10:43
flash blade environments consisting of a mix of high performance blades and chassis and lower performance capacity optimized expansion chassis. We will use the S 500 for the performance storage class and add expansion chassis to provide capacity. This way, the customers get the benefit of very high performance and and of course,
11:03
the TCO benefits via the expansion chassis that makes sense. So will flash blade with zero move tearing be different from other flash blade installations? Not really, like I said, from a purity FP perspective systems configured with zero move tearing will be powered by the same purity operating system. Uh And we'll have all the same software features uh from a F uh file or in future
11:30
object uh capabilities. All your data services like snapshots or applications save more. Well, all of those things will be uh uh possible. You will also have things like always on Q OS for the hot uh or the uh the file systems that are tagged as uh pinned as hot. Uh And you'll also have data reduction uh capabilities as well.
11:51
Uh You know, we pride ourselves in ensuring that we are delivering identical user experience across multiple product offerings and you should expect no less from uh systems that are configured with zero move tearing. However, I will point out one very specific thing that people should know zero, move tearing will be offered in very, very specific uh configurations.
12:13
Now, let me show you those a flash plate ZMT cluster configurations will be in these ratios. Uh you know, we will offer green field only and you will have the ratios of, you know, if you take a look at the, the chassis ratio S 500 with ex uh there is a one is to one ratio, which means there are approximately 1.5 petabytes of hot capacity or
12:36
three petabytes of cold capacity totaling about 4.5 petabytes and so on and so forth. The reason for doing this is to make sure that we are able to land the concept uh and prove it to be successful before considering other permutations and combinations. I also want to point out that in the first initial launch, this will be Greenfield only, which means these will, we will only support newer deployments.
12:58
You cannot take an existing S 500 add expansion chassis to it that would perhaps come later in the future. Uh But no decisions have been made on that. So for now, Greenfield only excellent. So when we think about, you know, the challenges that our customers are facing, what are the types of data that can benefit from zero,
13:20
moved tearing? That's an excellent question. JP uh There are multiple workloads that need the combination of high performance. Uh You, you know, for a small subset of the data along with uh you know, uh you know, keeping the data sets uh around uh for future accesses. Uh I can give you three examples. The first,
13:42
first one is ED A and chip design type workloads here. Uh uh You know, many of our customers in this particular space need access to a subset of data for doing their simulations, uh you know, and scratch space and they require extremely high performance both metadata. Uh IOPS performance as well as throughput, uh you know, but at the same time, they need to keep their assets around uh for future access,
14:09
uh zero move tearing uh with flash blade is a great uh great uh you know, fit for that kind of use case. Uh The second example is A I where you have uh the need for running uh a lot of high performance uh you know, training data sets, but you also want to keep uh all the data around uh in just in case you want to access it.
14:30
Uh And of course, uh that also means that data analytics uh use cases which require very similar uh high performance computation for small data sets. But want to keep things around uh for for future access for future analytics uh will also benefit from these. So these are three great examples of workloads that will benefit from zero mode.
14:51
So thanks Karthik, what else about zero? Move tearing our customers and our field team likely eager to know about, you know, JP uh whenever you launch a new platform and a new capability and a product, uh There is always questions about things like performance and pricing. Uh You know, I I want to assure our, our, our listeners that uh as we get, you know, as part of our launch,
15:16
those will be shared, uh you know, at the appropriate time uh pretty soon. Uh So those are the types of top questions that I get. Excellent. Well, thank you so much Karthik for joining us. Uh This has been really informative to understand a little bit more about zero move tearing and how we expect it to uh perform in the market with our customers.
15:37
So now let's pivot over to real time enterprise file. Uh We have John Carnes with us and John, you told me the real time enterprise file does for data. What pure storage has always done for block? What did you mean by that? So I love this question. Um You know, pure has been serving block data
15:57
for over 15 years now and customers have fell in love with it. It's the top block solution in the world, but many customers wanted that same experience, simplicity, non disruptive evergreen experience on file. And that's what real time file, real time enterprise file is delivering to our customers across the entire pure platform now.
16:18
So what is real time enterprise file? It's good. It's a good question. So it's it's actually um file on top of the pure platform. It's a new offering there where we take both scale up and scale out and give customers a true um you know, dynamic solution for files. This means we get around the limitations of
16:38
traditional file means you don't have to have these rigid architectures, you're not tied into certain uh um walls when you hit file system limits and things like that. It's a, it's a file solution that can change, adapt and reconfigure in real time. And this is something customers need on modern applications today. Well, absolutely. It sounds incredible.
16:59
So what are the core capabilities of real time enterprise file? So it's real core is being able to supply uh file as a service both scale up and scale out across the pure platform. And we're adding a couple new uh features into that. This launch. One of those is zero move tearing that Karthik has been talking about,
17:19
which is allowing us to do, get rid of the traditional tearing um cost in overhead and bringing in the same TCO or a better TCO. The other one I ne next one is secure multi tenancy for file. This allows our customers across the entire pure platform to securely um service access to file data out of both flash rain and flash blade.
17:44
So this means they can present unique name spaces um for their data across both products. Next feature that we're releasing is always on quality of service. This means that on both products um for file that no workload is left behind. If indeed there is ever a need, we make sure everybody gets the resources, they need to be able to make sure there's no downtime.
18:07
And then lastly, we're introducing copilot for file. So this is a uh anybody has seen copilot during the last accelerate uh for block, this is an extension that coil to file, allowing customers to go and use natural language questions to get insights, details and and help manage and improve their cost of ownership by owning pure with file storage.
18:32
So super interesting how, how does that differentiate from traditional file storage systems? So this is a very good question as well. Um Traditional file storage are built on technology from 20 years ago, many of them and this locked a lot of customers into these very rigid architectures which the vendors couldn't escape.
18:52
Um They had to, you know, keep their operating systems and built that around that rigid architecture. And what real time enterprise file does is it brings the cloud agility to that world and stop saying, oh, you're stuck in this world and you have no flexibility to change how you operate and improve your cost of ownership.
19:12
Real time enterprise files brings all of that to our customers. It allows you to not have to preplan for future gross and con consumption. It allows you to dynamically change unlimited uh your file system size and be able to continue to grow as your data goes. It, it allows you to take advantage of all the goodness that comes with pure with the denis uh direct flash modules in the in the industry.
19:38
So when we, when we put our customer hat on and we think about it in terms of how this you know, is likely to be used in, in the wild, what are the benefits of using real time enterprise file, there's several, I'll try to go through them one at a time and, and give a little explanation. So first is dynamic adaptability.
19:59
So this means that file services can adapt as you need it to real time. You're not gonna have to uh reorganize your data, reconstruct it do migrations to be able to uh support larger workloads or changes in your workload. You can just do that whenever you need to. Um you can also upgrade your hardware non disruptively.
20:20
Secondly, simplified management. What this means is that you get a real cloud like solution. This includes things like fusion for pure, which is where you can actually do um operations across the entire fleet of the pure platform. This means you can uh dynamically change and deploy workloads across any of our scale out
20:43
scale up file solutions. Thirdly enhanced security because we have the Secure Multi Tennessee and we have comprehensive security for that. That means you're actually able to take another step forward in this world where you know, cyber security is so important, you can take another step forward in that in your file workloads,
21:03
improve performance because we have that always on Q OS running all the time without any impact to your day. You know, you're not taking a trade off. We make sure that your most important workloads always have resources to be able to go and operate. So we understand any downtime is costly. We don't want you to have downtime.
21:22
And then lastly, obviously, as I've been talking about all these is cost efficiency, zero, moved tearing um with the removable of all the complexities and in management overhead can dramatically increase or decrease your total cost of ownership of tarring environments such as ed A workloads that uh Kartik talked about. I think these are the prime areas that customers can take advantage of.
21:45
A lot of these places is where customers see issues today. And we wanna make sure that they're all able to remove these uh factors out of their life forever. John, it sounds like there's a lot of benefit for customers. And I have to ask who's gonna benefit the most from real time enterprise file.
22:04
It's really anybody that's kind of locked into these complex, you know, architectures and, and probably the most is the ones who are losing their um storage in staff where they have a small group of people trying to manage a large large fleet of data you bring in real time enterprise file. And that means all their, you know, file and structure data is super simple to operate and manage from now on with
22:27
the with, with uh pure fusion with both scallop and scale on the same uh solution. You can dynamically do all that and not have to have a huge team to manage it. It's very easy to learn and it's very easy to continuously take advantage of. So this seems very innovative. Are there similar offerings on the market? So are there file solutions on the market? Yes, there are,
22:54
I mean, obviously, you know, competitors like Dell netapp HP, they all offer some file file solution. But real time enterprise file is unique in its ability to leverage the pure platform. And this provides unmatched real time flexibility, simplified management, enhanced security. This is something that we do not see and we've
23:13
been asked for by many of these customers from the competitors. It's super exciting to see the evolution of the pure platform. And Karthik John, I'd like to thank you both for your time today to give us a bit of a deeper look at these incredible new features uh to our audience. Thank you so much for joining for your time commitment and your continued interest in how
23:35
pure storage is creating the storage platform for the next decade. Have an amazing day.