00:03
Hello and welcome to the session of next gen connected cloud data center with programmable and flexible infrastructure. My name is Jack Hogan. I am the vice president of technology strategy for pure storage and I focus on our cloud edge and service delivery partners. I'm a former 20 years CTO on the customer side, five year pure customer and I've now been at
00:28
pure for 4.5 years working on our partnerships with Key Cloud edge and service delivery partners. I'm joined today by Prakash chandra Prakash. Why don't you go ahead and let everyone know who you are? Thank you chad. My name is Bikash Roy chowdhury. I'm a technical director focused on h B C E D, a software DEV and DEVOPS.
00:48
That's quite a mouthful, but that's what I'm doing. I'm four years old and pure and going strong for primarily focused on a lot of new architectures in the cloud and um, scalable models that we can work on on our flash array and flashlight models. Excellent Prakash, thank you so much. So here's what we're going to take you through
01:09
today. We're gonna start off by talking about bare metal as a service, bare metal as a service is really the foundation for how we're delivering our connected cloud storage solutions. And then we'll go into what pure storage for connected cloud can do in that will go through the connected cloud validated architectures and specifically focus on what we're doing with
01:31
Porter works and G K E anthros on bare metal and then go into what we're doing with Flash Blade connected to the Azure compute for scale out HPC workloads. Then we'll close out with key takeaways from all of the elements that we're going to cover. So let's go ahead and dive into the content, starting off with what bare metal as a service is and how it's ultimately solving problems for businesses and what we're delivering today at
01:57
the edge. So we first of all want to start by talking about what most companies are facing today in their cloud journey. Now this cloud journey may not be ultimately landing into a public cloud hyper scale or it may just be moving towards digital transformation to advance company's ability to be more flexible and react to business changes faster as well as have higher order
02:20
capabilities that the cloud promises and delivers on. But when you start with this and companies realize where they are in the journey, you have to really understand what the past challenges, the present challenges and the future challenges that face any organ organizations I. T. Team. So past challenges we're gonna find are going
02:40
to be a lot of your legacy on premises applications, those applications that were never architecture or designed to run in cloud native principles. Oftentimes these are large scale database workloads are large scale, file scale out solutions or those that require bare metal server access, even virtualized machines that really aren't natively portable into a public cloud construct,
03:05
but of course every business operator in I. T. Needs to be dealing with their present challenges. And those present challenges are the growth of data, high volume data driven applications driven things by by things like data warehouses and scale out unstructured data that's coming in multiple different forms, even streaming analytics and of course the intelligence and Ai and machine learning that
03:27
you want to put on top of that data. And of course then most companies want to get into that future state to be able to allow for the benefits of cloud native and next generation application sets, leveraging solutions like containers or kubernetes using cloud native services delivering multi cloud engagement. So accessing all the different clouds and the
03:50
solutions that they have and of course being ready to handle next generation opportunities. Um And technologies so really as you look at what we see and I talked to a lot of uh a lot of C. T. O. S. And a lot of I. T. Leaders and the challenges that they're facing today in this journey to cloud. Well first of all they usually are force was dealing with that designing building and
04:13
managing everything on their own. And in many cases applications that weren't meant for public cloud or improper sizing or going to lead to massive cost overruns. Uh And the the legacy technology challenges the ability to actually pour applications directly from on premises infrastructure to a cloud architecture. Oftentimes requires a lot of re factoring and expensive retooling to ultimately get
04:42
to achieve the best the benefits of a public cloud. But the real challenge is that most companies, they really lack the skills to get their traditional I. T. Operations. Teams are not equipped to being able to deliver on develops principles and retooling and retraining these folks is really creating a challenge that uh that slows down the adoption and really puts puts businesses at risk in
05:08
terms of making the jump straight into a public cloud. But what Pierre is doing here is delivering a bare metal as a service capability. Um that bare metal as a service capability allows for high performance on demand hybrid data platform. And it delivers the same thing as a cloud would but without those risks and costs of the hyper scholars because it allows for the same
05:31
infrastructure that you have in your on premises data center, the same levels of control, same levels of data sovereignty and regulatory compliance, same levels of understanding what your teams can do but all of those can be delivered in a cloud like fashion. So let's dig a little deeper in terms of what bare metal is, the services and and what it's really delivering.
05:51
So first of all when we look at solving an uh an outcome or delivering an outcome to the business with I. T. You really need to look at the entire stack. So starting with where your infrastructure is placed moving up into how it's interconnected, how it's actually physically managed and racked and then ultimately automated above there and then you get into the software layer and then above that is really your end consumer or user
06:17
layer where the data itself is generated now what bare metal is a services is really the hosted hardware as a service. But that gives you a very different experience than you get into a public cloud. Whereas a public cloud or infrastructure as a service is typically a multi tenant environment that takes away a lot of the controls that you have in in an unprejudiced world, what bare metal does to solve that problem is they don't
06:44
get delivered physical single tenanted servers, network and storage that are only for your consumption, meaning you're not on shared systems, you're not having to rely on proprietary control planes to virtualized instances. This is the physical bare metal where you have root level administrative access, but it's still able to be delivered with the same cloud principles that cloud like ease and
07:10
speed leveraging A P. I is to stand up physical infrastructure or leveraging cloud control or web control planes to be able to deploy net new infrastructure in a matter of minutes. And that's where up stack solutions like we're gonna get into here in a in a bit um can be delivered on top of that. But with all the benefits that you have on your
07:31
your current on premises or in your own data center infrastructure. So if you realize now, what this can deliver is really that ability to take all of your challenges, the past, present and future challenges all of your application types, all of your key business drivers that are driving data as well as delivering cloud native capabilities into a single platform now appears partnered with Equinox in this Equinox
07:59
metal is the infrastructure capability that they deliver. Pure storage is a primary provider for storage there and so we can provide our solutions to unify all of the capabilities that we've talked about to this point in one single platform from one single vendor allowing for all those unified capabilities, all that unified connectivity into the public clouds.
08:23
And note that providers like Equinox actually have hundreds of thousands of interconnects allowing you direct on ramps to public cloud as well as to your own data centers, allowing you to really stretch the architectures from your current on premises environments to your future cloud environments, all in a single, highly secure automated interconnected hybrid solution in a single platform and
08:49
it delivers all of this infrastructure at the speed of software. So as we move from here, because why don't you go ahead and talk about why this is actually solving and what we're doing with this type of architecture and the solutions were delivering from that, Thank you Jack. And the big question why? Now, all the great stuff that you just heard
09:10
from Jack talking about bare metal is a service and all the integrations and and automation that we've been putting together a place. What is what is so important at the timing And while you're thinking about it right now, first and foremost is the data growth. Now look into any vertical, any solution segment, any business processes that you do see in practical world today, everything and anything that runs software generates a lot of
09:38
data. There is an overload of data, huge volumes of data that second is all about the model. Okay you have this data, you just can't throw away the data so you like to protect the data to run analytics on it. You do a lot of different data management operations on top of that. So how do you actually enrich yourself on the data management experience that you are trying
10:01
to go with and that is where your story shines and we're going to talk about more as we move forward apart from the data volume and the new modern day to experience what else that is the cloud. Now we already have customers already have multiple solutions and multiple options and choices to make to set up the application in the cloud that is born in the cloud. That is one way of looking at it.
10:26
But there are a majority of the enterprise customers today who are running on traditional workloads and the data is absolutely critical to them because when you move to cloud, you hardly know where the data lands is it going to be in the US West US East US West based company. You would love to have a lot more proximity to the data, but apparently you have no clue whether it ends up because you could be sitting
10:49
somewhere in a in a PJ for the modern asia, who knows. So the control over the data is absolutely important for most of the major enterprise accounts because data security and sovereignty and that is something which we'll talk more about as, as we move forward. So if we do not start exploring these areas now, now we lead to a data,
11:11
a loss of business opportunity. That means we are missing our big time in this particular space where we need to start transitioning with the customers. So why connected cloud? Now we can say, okay, we already have storage vendors already there in major cloud providers.
11:30
The hyper scholars. So what is pure, what is pure suggesting to go with the connected cloud and what is connected cloud, basically connected cloud is basically connecting to any cloud provider. That's the beauty as the term says multi cloud. So if you choose to go with a single storage vendor and the vendor could be already tying our underpinning inside a major hyper scalar organization, then the
11:55
choice of moving data across different clouds is becoming a little more cumbersome and challenging at some point of time. So now if you were to give you an architecture which gives you a lot more fluidity, more flexibility, not only to stand up the entire infrastructure on demand, but also the ability to move your data have control over your data.
12:17
So the data sublimity is very important. The flexibility of setting up an architecture or platform is equally important and bursting in the cloud. Now, when you burst into cloud, I can tell you that okay, we have got an infrastructure is up and running, it is flexible, you've got a lot of the automation in place, everything is done and what happens to the
12:36
application itself? What happens with the application? A licensing model is not capable to run in the cloud. What happens if the schedule or that your running to run your job's on premise doesn't have the ability to scale in the cloud. So a lot of these pieces tying together to make this business operations successful.
12:55
So when we go and validate an architecture like that, we make sure the applications were running has a licensing model that supports cloud, has an ability to actually have the schedule is to scale to the cloud seamlessly. So that is how we actually run this architecture and test, which we'll be talking about in a minute where we will be later in this presentation where we actually do do to see the entire stack and not just the
13:20
infrastructure itself? Yes, infrastructure offering is a crucial piece and a big piece of it. It's a huge heavy lifting but it's all been taken care about your storage and economics as we move forward in this presentation. And then finally the important part is our next model because you don't own any of those hardware, you just pay what you use, you just pay what you consume.
13:39
That's very important. So now, if you look at this diagram, let's start from the bottom left. That is the traditional data centers that we used to have over the years, I don't know for decades right now. Right. But the challenge is more and more of those data centers are starting to consolidate,
14:00
the business is starting to consolidate from lots of many number of data centers they had over the globe into a very specific and and a small amount of data centers that can be managed from a from a Capex perspective. But what happens does that mean that they're going out of business now? They're actually moving the business to a more elastic and flexible model and for that is the
14:23
reason why we are we think the timing is great because then these things are happening considering the amount of form factor that we have in our appliances today, how it actually fits into the rack space of a of a cloud provider or a co location like Equinix and they're in direct space and then how do we tie into the network connectivity to any number of cloud providers that we can possibly have that obviously we're looking at
14:46
the three major ones in this presentation but we always have options to, it's for others to now if you look at the center and that is where we are trying targeting to head the most because here you get the benefits of the appliance, the performance that is needed. So you're not compromising on the performance, but at the same time you're also making it an easy way of consuming storage capacity and
15:10
scaling with respect to performance and capacity seamlessly on demand. And then that becomes a lot more useful and powerful for a business because they don't have to care about the underlying infrastructure and their segments that they need for those applications, all they need to care about what the business requirements are and what the outcome says. And then if you extend that beyond,
15:32
if you look at the from the middle which is the co location and then you look at the right hand side where you extend that to any cloud provider. So that's the that's the major advantage that we're providing with this architecture, that is why we call it as the next gem flexible architecture. Because first of all, standing up the infrastructure as Jack was pointing out earlier,
15:54
there's a lot more flexible because of the automation that we are providing and the validation that has gone behind it to support that. And second is that you have got the entire choice of moving your data to any cloud provider depending upon the cost that you look for. So for the purpose of this presentation we will be focusing on Azure because we picked up one
16:13
of the HPC workloads as well as um we have also used google cloud for port works. The two different dis similar kind of a platform. We choose to give you a variety of what we can possibly do and not try to be completely repetitive in my on our testing so that can give us a little more knowledge and insight into this architecture which we can play as an edge as Jack.
16:37
Jack was introducing himself as an edge player. I would like to continue on the data point that we are all going to be an edge player but we have got various different business requirements and various different kinds of platforms to play with and that is where we actually add value to the various different layers in the portfolio that pure us to offer to integrate into the cloud in a habit mode.
17:01
Now what are the benefits now we've looked into why was it important and how will the new architecture look like and why? What is the benefit that we are actually providing? What is the value basically I would say so first of all is all about secure data ownership. That's very important because data is the new oil and that has to be secured that has to be protected.
17:24
That has to be preserved. When I say preserve means long term retention archiving you go through a lot of your audit requires in the United States or in the media for any other regions of the world. Data is important. Second is do I really have invaded to augment cloud? Supposing I'm running low on my data from the space.
17:44
I want to move away from the complex during offense and even though you can do an optics model which we are offering today on your traditional data centers. But what is the, what do I have an option to move into cloud? So that is where we play a major role as to how do we tie this entire experience to our customers with the customers? So that will be a lot more single second is
18:04
okay. We got the cloud setup done now. We know really the application can really work because think about this, you've got compute resources running in the cloud, you have got a storage infrastructure sitting in a co location, there is an express route or a direct connect connecting to that particular um infrastructure. Um Equinox there is a 2 to 4 to six millisecond
18:26
round trip time. Will the applications ever handle that kind of latency? Because you're on your own premise. It used to be seven milliseconds Less than one militant. So when we move into cloud. Those are the things the questions we always asked. Does the application really handled, Can it
18:42
tolerate that much amount of latency? So that's the test that we are planning to do and we did to make sure that our we really right and making ourselves corrected as the direction of the path we are heading right now is the right way to do it. And finally yes, we got the performance also out of the way and how is it effective? Is the cost model? Is that something that is really worth
19:01
expecting of spending that much money and how much it has cost me at the end of the day. So those are the things we always start to think about it and that is where I think we play a major role from a positioning perspective in the portfolio that pure storage has today and what we can offer by integrating with the clinics and other cloud providers that we have and the applications running on it that uses a state fel storage behind the scenes to
19:26
make the data persistent. That is what we're going to talk about. So the first validation we did was with Equinix Metal and board works. Now look at this schematic, you're right on the top left. You see, manage services. The managed service is all about community services would be um, AWS EKS or as you're a Ks or
19:53
google G K E so they're all managed services and the reason why I have the unmanaged service on the right hand side Is that it's something that we can still do it only critics metal you don't have to be only on Manus services, like the purpose of our validation. As I said, we're touching about two extreme use cases and with the extreme not use case, I would say about extreme platforms where we're trying to choose to go with the communities
20:17
waste container platform and a VM based, a typical file system which is a non container, non communities environment. To see how do we hold up in the entire state from a performance perspective. Now there are a lot of different architectures we can choose to work on if you look in here, we use google kubernetes engine with Anthony's which is the control plane from google, this is installing Equinix metal which has got pork
20:44
works, underlined the covers As the data management layer. Now this entire stack has been automated using terra form so you get the customization and the flexibility to stand up this entire environment just under 15, 20 minutes. Is that easy. Now, what we're trying to do here is we're trying to say,
21:00
okay, we got this infrastructure. Now let's try to see how we can compare it in the various different flavors of the implementation. Now a customer choose to go, okay, I've got a G K set up on a local local SSD storage, which is what typically what you do when you start to spin up a geek, a cluster.
21:19
then we choose to go, okay, let's use sport works as a data management layer. Right? And then finally in the middle we have got another combination. We're using Antos, google Antos on equal X medal with a broken self, you probably know Rockin Chef specifically Brooke is a C N C F project, which is now um um it's not in the sandbox anymore,
21:42
it is getting ready released to be in production. So Rook is any more and more popular with a lot of users but the look is great. I don't, I'm not going to be dismissing the fact that it is not great, it's really great but it lacks a little bit more challenges of features, functionality that is needed for data management.
21:58
And finally on the right hand side is that what the architecture we chose to test and that is having all the answers and google engine and kubernetes engine in there on Eclipse Metal with support works and as I said, this is all automated. So there's four different architectures that you see in your screen right now, Iran and Y C over the Mongo dB testing and um on, on this,
22:20
on each of these architectures and we try to measure the throughput, the latency, comparing that with our extreme bright architecture, which is what we are proposing today and try to do the comparison as to how well and good we performed against other three architecture that you see on the screen right here. Now to summarize this, What we found if you see here the anthology
22:44
e m stands for Equinix metal and PXS sport works. You see in each of these top four categories, we had a three X improvement in throughput compared to the native google cloud. At the G K cloud geek. A cluster in the google cloud, We had a three x latency improvement in the GK. Google Cloud,
23:06
25% throughput improvement with panthers and rook. And then we had a 30% latest improvement on and thorns and drugs. So what I'm trying to say here is for a database which is running in this infrastructure. We are actually beating the traditional formats or the layout that we normally see customers doing going with G K clusters or even using um
23:34
support works for that matter. Straight photo. But if you put that in a bare metal on in the iniquitous platform that we're still getting a huge amount of benefits and the performances itself. Right Then the last one is all about the flexibility of setting up this environment. We were able to create this entire environment from ground zero up under 40 minutes.
23:56
And that has all been configured and consumed by terra form is absolutely flexible and that's the reason why we call this as a flexible architecture, which is going to be the next generation platform for a lot of the more on applications that has been done to the generator. So apart from the performance, what else I just mentioned earlier? Data management includes data protection to how
24:20
do we protect the data now? Not only the protecting the data and also it was the mobility to multi cloud. Like if I start off with the EKS cluster like in this example on the testing that we did, we started with now if I do have a geek, a cluster of my data sitting up there and tomorrow in the future I choose to move into a different cloud provider. Can I move my application?
24:40
Absolutely. You can move your application, you can also move the data along with the application and that is exactly what what works is capable of doing and because both sets the layer which can actually manage the data irrespective of what kind of clouds, what are you having that gives you a common platform which can extend to any cloud water and so does your data can move back and forth between the cloud providers to
25:06
So as I said, we were testing to super set of platforms. The other platform was primarily testing for traditional applications like HPC applications and specifically in this one is E D A E D A stands for electronic design automation, which is mostly the high chip design companies where like the AMG is the intel support comes the Broadcom's, you just named them all of the chip design companies,
25:27
not the manufacturer, the manufacturer also falls into that, but that is mostly E D A stands for the immediate tools that have been used. And those tool vendors are primarily synopsis, cadence mentor graphics whose tools have been extremely popular in the market from the, in the semiconductor chip design. So we chose to go with Azure as a cloud provider for this validation doesn't mean that
25:48
we cannot run any HPC application for with AWS or with google cloud absolutely as possible because the underlying construct of this structure is again sitting in the economics location. So if you're sitting already in an Equinix location, you can connect through the clinic's fabric as you see right here to any any cloud provider in this example, you're seeing only Azure and it said you can replace that with AWS
26:14
or will be cloud for the matter. It is just that the purpose of this presentation, we have got maximum validation with Azure and as sure as a problem, partner has been extremely supportive and and has been working very closely with us at your story. So that's the reason why we're able to roll out so many solutions in a very short time.
26:33
So if you look at this entire layout on the left hand side, you see your traditional data center and what you see in the middle is where you actually extend or stretch your native data center to a more cloud adjacent cloud connected model through an Equinix fabric which can connect to any cloud provider and you can actually provide that in the different regions will be having our offerings in any different
26:56
regions in different countries. So that is a completely global offering. It is not tied to us, only in us, it is completely global, it is not tied to only E D A. It could be any HPC on the analytics workloads moving forward, it is just the sky's the limit, a number of opportunities that you can envision through this architecture that is possible.
27:22
So for this validation, what we did was we took a flash plate sitting in an on premise which you see on the left hand side and then we set up this entire stat of flash plated and Equinix location tied to an as your compute which is connected to an express route with the two millisecond round trip time. So that's very important. So when we started this project and it was mentioned that as to the benefits of the new
27:44
slide is too, does it really help us because yeah, we know flash braid works great and on premise and sub millisecond agencies and well below one municipal agencies, the applications are really tolerant towards that and we do a fantastic job in handling those work clothes. What happens when we actually moved to a two millisecond round trip time with the Azure connected gateway um like your ultra
28:06
performance keeping you see on here. But those were the different apprehensions we had in the beginning. And that's the reason why we chose to run real media workloads. And that is thanks to Azure and the collaboration to it. They have the tools vendors like synopsis where we actually run those kind of applications and
28:23
real data and in the in this hybrid cloud model to see how performance, how flash that can handle those performances around the crime. But the amazing thing is We are we this is something which we just ran and we ran during this validation was a spec storage 2020. This is an industry standard benchmarking too, which is kind of multiple profiles for various different workloads.
28:45
And we chose to run the media profile and this is something which we are comparing with. Um as you need to fast for a document which has been publicly available and the link is also there. So what we all saw, if you look at the left graph, you see the orange line moving all the way towards the right and starting to go up. That is the flash plate.
29:09
So that is where you see the linear reality. Now the blue line is the N. F. Now considering N. F. Is already inside the agitator center which is actually around 0.8 millisecond round trip time compared to two milliseconds for us. We start a little higher with a little more higher than one millisecond in the initial
29:26
stages. But look at the linearity that we have provided a until we hit the hockey stick curve moving forward. And at that point we saturate the express route which was a 10 gigabit per second, which is a one gigabyte per second throughput. And that is the limitation that we hit as soon as we hit that with our curve starts to go up And that is a clear indicator that either we
29:47
can start growing the network bandwidth from 10 to 20 to 30 to 40 depending upon your requirement. And you can still see the linearity. Now the good thing about this is this flash blade, which is a three chassis, fully populated 45 plate Was hardly doing much, it was like 25-30% utilized. But when we were trying to stress it out and trying to scale the workload so compared to the
30:09
N. F, we've been outside the data center with two millisecond round two prime. We are still linear and we're still able to deliver much higher throughput and bandwidth. The better and better agency compared to what a NF can do today. On the left hand side, there is all over NFS V three because media workers are primarily and
30:27
predominantly our NFS V three. Now there is a there's a part of the media workers also run over NFS version 41 for security purposes and authentication purposes where conversion for one is not basically a performance play. And if you look at the right hand side graph, the flashlight actually has got about 85% improvement over a N F, which you see the blue curve just going up right there.
30:52
But we are still live here and keep going onto the right before we start hitting going up. Now, I can clearly say that NFS version 41 is not a parity with benefits portion three from a performance perspective and that is not what it has been designed to do. But the the key differentiator is how we can handle NFS version 41 in a flash played versus what we can do with the N. F. So that is the key difference now
31:19
just to summarize in the cost aspect of it and that's the thing which we need to cover because we talked about um we talked about the flexible architecture, we validated the performance um and making sure that the application is actually really performed well based on what the requirements are. And the third thing that is very important thing is is it really worth the time and the and the cost and what is the cost associated
31:41
for this entire infrastructure. So first of all, I can give you the cost analysis that we did. It was up to 2.66. Now, what is that compared to that is compared to the N. F is capable of doing today and we are up to 6.2 point 66 with the 10 gigabit per second. Um extra strap as you keep on increasing that
31:59
to 20 to 30 to 40 but we are still about 2.3 X. You see on the second box is a 2.3 X cheaper. So we're still much better from a cost perspective as you start to scale but you're still getting all the benefits of performance, you're getting the feature functionality that you need to manage your infrastructure plus the automation part that is needed and the ability to scale the computer notes in the Azure on
32:25
demand when you start to burst into clouds. So all of these things all packaged together and you're still getting this a lot more cheaper and then what exactly I was talking about is the egress costs and there's hardly anything or almost zero egress cost dash and that is another big attraction and the entire setup as I said is completely going to be automated and it will be set up from scratch. So when you see so many players over there is
32:48
pure stories, there is uh Equinox involved as as you know who I talked to. So that is always a confusion and confusion and ambiguity that comes up in mind but I think we have already been taken care of all of that and in the future for the conversation please reach out to Jack Hogan um at pier storage dot com for more confirmation as to what exactly needs to be done and Jack and pointed in the right direction as to what we are offering today from
33:11
a, from a infrastructure perspective and who will be the person to contact and also wants the infrastructure up and down. Who is maintaining it and who do I actually call if I'm having challenges and all of that stuff. So everything is all taken care of and rolled into one umbrella. So it becomes a lot more easier way of accessing this entire infrastructure and
33:28
consuming it as you start to scale your um workloads beyond your data central boundaries and finally all the benefits of flashlight that you get everything, your performance, your feature functionality. So nothing is that your data reduction. That's huge because when you get the data reduction on your on premise flash plate you'll be getting the same data reduction on your flashlight which is sitting in an equity
33:52
location because the hardware is no different. It's the same hardware. So basically you will be getting the same amount of benefits, benefits and efficiency that you get from a flashlight on sitting on our on premise location and finally the ticket base just to wrap up what exactly are we does this cloud connected claridges model really help.
34:17
So it's all about architectures. As I said we were trying to quantify or qualify or validate two different kind of platforms. Either use it for tier one, Tier two. What kind of water, What type of applications you would like to run databases in a container or bare metal or you use the cloud for that matter. But underlying infrastructure is a the physical
34:39
hardware, like a flatbread sitting in an aquatics location. So it is as I said the capabilities and and the combinations are are many and you can think about any workload. We can actually accommodate that. Yes we can. And for the purpose of this presentation again I would like to repeat that this we have selected certain workloads for the validation
34:57
but that does not stop us to validate any other business requirements for either container based or on a file based which is a non container non competitive environment. Second is a scalable performance. That's huge because a lot of the performance are is buried or is driven through cost like okay I'm trying to add another 1000 cores in the cloud but to support the amount of I.
35:27
O. And the traffic it is just not the network conduit connecting to the back and storage but also the storage capacity also has to be equally bigger because that is how the storage is going to handle the amount of requirements for the computer ship provisions but here we are completely doing this independently. Now as you start to scale the number of course as I said The flash played we were literally
35:50
like 25-30% capacity when the network situation happened but that are you paying for this entire flash plate? No, you only pay for whatever you have consumed. So that that is where the cost becomes a lot more more competitive and more what you call attractive from, from a usage perspective because that is the reason why we actually have this architecture in this place and the family addressing the concerns about data in data and
36:15
cloud security, governance, everything all put together. Yes, that is your, your space, your data, you know where the data lands, you have got full control over it and setting up the entire infrastructure is completely flexible. And please reach out to Jack open for getting more and more information about how do we actually go about talking to economics or any
36:37
cloud for a provider for the matter and make it a lot more use user friendly and easy experience to set up this entire infrastructure for your any kind of a business worker that you would like to plan to work on this uh, on this platform and this cloud connected platform. Well gosh, thank you for that. Thank you for running through the validated designs. I want to thank you for everybody listening
36:57
today as Bakhash said, you can reach out to me personally Jack at pure storage dot com. Or you can visit www dot pure storage dot com slash bare metal. All one word www slash www dot pure storage dot com slash bare metal. To learn more about the solutions reach out to those that can help you and ultimately achieve new flexible cloud infrastructure with connected cloud architectures.
37:25
Thank you all. Thank you, Bakhash and all have a great day.
