 Good afternoon ladies and gentlemen, so welcome back to the track. We have a great presentation for you We have Tim over here who's gonna lead a panel with a number of customers Pretty interesting perspective there. If you weren't here this morning a few things I'd like to share with you We have the HP Sponsored community lounge. It's over back that way. I haven't made it out there yet today I highly recommend that you do we have some great hoodies with iron-on badges, which is so you can customize them pretty cool I also highly recommend that you RSVP for our party on Tuesday evening. It's gonna be a great time And it's gonna run out of space very soon. So we're there further ado. I'd like to hand it over to Tim Nice Cody Well, welcome welcome to OpenStack and to the sessions here with Hewlett Packard I'm with Hewlett Packard. I'm in the product management team and I have the Honor of working with many of our customers. I run a program That talks to customers on a regular basis for our get customer requirements So for today, we brought together four of our customers I'll start on the side over there. We have Steve Duft with HP IT and I know it's not fair because he's Hewlett Packard But if you've ever seen the inside of HP's IT group, they're a customer. They're one of our best customers They're one of our biggest customers. So Steve and I have a great relationship that he's a customer to me So he's on stage for us today with that We have Orlando Bader who is the CEO and founder of Orimucco And he they've been running OpenStack for quite some time. We've got Devon Bleak with Fox Entertainment And we've got Scotty Miller who's here from DreamWorks. So we got a couple on the end here representing the entertainment industry. So What I'd like to do is have them each tell you a little bit about what they're doing with OpenStack today in their companies So every one of these guys here is working with OpenStack today some in production some in in development and test and I'll start with Scotty, I'll have you start off with DreamWorks kind of what how you're using OpenStack today Hello everybody our use of OpenStack started Probably in 2002 before cloud was cool back then it was called utility computing or grid computing and it was a way to Share infrastructure as a service resources that HP had in their Palo Alto data center that we were using in our Redwood Shores Facility from that relationship. We move forward to using HP's public cloud for object storage Using the cloud primarily as a way to store cold data But also using what I call cloud as transport you put the data in one side and it pops out the other side and someone else's Network transits your stuff around the globe pretty pretty handy We've been doing OpenStack itself In the DevTest environment for a little over a year and a half now and we're just about to receive our second OpenStack cluster We're gonna be a Customer shipment number one of the HP Helion rack product, which will be kind of cool It's a control plane in a box with enough storage and compute to be useful and we're gonna use that as our DevTest and early prototype for Some outward-facing image generation media delivery services projects that we're working on as well as Rehosting our internal cloud to be OpenStack instead of the human middleware driven telephone and ticketing system cloud we have today I think you're mic. You're mic'd up. Oh so We've been looking at OpenStack primarily along two kind of tracks The first is as repatriation from some of the public clouds of some of our applications and the other is to help enable more agility for some of our development groups internally in the enterprise Perfect. So for those here that are not familiarized with Ormuko. We're a service provider telecommunication Managed across the road. So basically we decide to use the OpenStack as Or the top 10 video game companies five of them are currently our customers and we have to host their games which have around 720 physical servers that we have to host whether in the Americas or EMEA and We decide to build a next generation cloud and we use OpenStack for it so We decide to use an opening standard that allows us to deliver the next generation cloud to run application that Existing public cloud providers were not able to provide Thanks, and Steve you're required to use HP Healy on OpenStack. Is that correct? technically no, but So we have been using OpenStack for a number of years obviously HP has their public cloud and we've been running applications in that space for for for quite some time we also About two years ago decided that we had a need for an object storage solution You know and we had what was there in the public cloud But we also had some use cases where we wanted it behind our firewall that we could use it for More more traditional applications, but having that class of storage that we don't have today so we just we deployed a Swift stack based OpenStack object store and You know since then we've been moving toward what was originally cloud OS from HP and now the the healing on product looking at The the full infrastructure is a service suite the the development platform as well really to to provide Platform for for application transformation for to allow our our app teams to develop a more cloud native form But run on our internal infrastructure Great. Thanks. I'm gonna I'm gonna pick on you Devin first So I know that you guys at Fox Entertainment so you've got all these movies that come out So you have all these third parties that build applications or websites to promote these movies? And you as your cloud is providing a service to that is that is that correct? Yes We basically built a platform within the cloud to all of these third party built applications So when you take a look when you look at OpenStack that being an OpenStack cloud What are the challenges in having that many clients coming into your data center and and providing either applications or? So for us, you know the cloud is all about agility of the infrastructure layer We don't know what our marketing groups want to do with their next campaign So we can't really you know say buy go buy a whole bunch of hardware for it Additionally, we don't know when that campaign launches how much traffic it's gonna get or how it's gonna perform on the server So being able to scale up Compute and storage kind of at a moment's notice as well as you know scale it back down when we don't need it again That's really what we're looking for and you guys are that you're running today on OpenStack So you're moving an existing workload to that looking to move it to an OpenStack cloud. Okay, great Orlando As a service provider You have extra challenges with many clients that would be coming into your cloud and I know we talked a little bit earlier and Jonathan talked this morning about Federation. I hear you're doing something around Federation with your cloud today. Is that right? Yeah, that's correct. So basically This is my advice for everyone here that is trying to build an OpenStack cloud that require more than a hundred notes There is many challenges company have nowadays. They want to become Service provider within the organizations and one of the major challenges They have when they build OpenStack is they cannot scale it to the limit they require to run their business So if you're planning to build it, you should be looking at a federated cloud So what this means is if you want to have more than a hundred notes You don't want to modify the OpenStack code. You don't want to go out of the standard You want to make sure that you can put 10 members which is Basically 10 OpenStack installation of a hundred notes to be able to go to a thousand notes I mean able to manage it, but it's not only to have a single sign-in. It's just the ability to Avoid all the performance that you actually have by running one single OpenStack installation for your organization Okay, and so you you challenge you Went after Federation and you have Federation working today as well. Yeah, correct. So we actually made a work It was a big challenge to make it work before with Juno like I'm going to tell you it was almost impossible actually without modifying the code I Don't suggest go that road if you go with Kilo We spent on 24 months of R&D to actually build the federated cloud around 31 employees multimillion-dollar project With Kilo, it was actually easier. It was not like they say it's already done. It's not production ready yet you need to actually look at SAML and Find a way to incorporate in your code because it's It's already built in keystone, but it's not going to work the way you want it and don't try to adapt No, don't try to change OpenStack itself Like adapt your code to work with OpenStack because if you change it You're not going to have support for maintenance releases and patches and we have been running in production So just my advice for anybody here is just keep it as simple as possible and don't try to Over-complicate it, right? Great. Thank you. Thank you now Steve HP today the one HP We'll be to HP in a little while, but the one HP right now has over 300,000 employees. Is that correct? Yeah, we're on there high 200 so so you you definitely have to deal with scale issues When it comes to the cloud so what are some of the challenges that you've had in with OpenStack on dealing with the scale issues internally? Well, so I mean it's it's always the same challenge. I mean you know the reference that You know that Orlando was making to Scalability limit limitations of a hundred nodes and whatnot those those always exist are we have an existing private cloud that runs on HP cloud service automation and a VM bar back end that we built over the last two or three years And that's running somewhere around 36,000 VMs and 1700 applications It's it's difficult to do that in environments where where you're constrained and so to his point building building federated Buckets to be able to run your applications within is a much cleaner model rather than trying to force scale out of out of the entire environment Yeah, so his point that it's actually right so for example Limitations if you're here is because you're interested in OpenStack If you're trying to build neutron, you know is one of the weakest thing that exists right now, so don't try to Make it that's you know, like high available Modify the code just just use hardware to actually connect to Neutron and make it work. That was the only way we actually were able to make it successful the same with the storage If you can only go to one petabyte Use federated environment to go to ten times that Right. Yeah, Scottie. You're not in your head. We had a little discussion earlier on networking So tell us a little bit of DreamWorks as challenges in the networking space and how you attack that So the biggest issues in the networking space for us is and It goes back to one of the social engineering challenges convincing developers and software architects to run in a cloud Architecture this whole concept of cloud ready has been a little bit of a challenge What I mean by that is most developers are used to very resilient highly Available physical infrastructure in which they run their application so they can be a little bit sloppy in a cloud environment It's more dynamic VMs can disappear hypervirus can disappear things can disappear and the application has to provide the resiliency So design applications to be cloud ready means the applications are restartable. They're recoverable So we run headlong into the need for resilient infrastructure with the single points of failure that are in neutron DVS has some challenges. There's other issues. We had with interaction with our routers Where a neutron has been probably the problem for everything we've tried to do in the last six months So again to what Orlando said the recommendation from some of the cloud architects is to do networking redundancy and physical infrastructure and Then maybe even look at using VLANs for your network subdivision instead of relying on neutron today It'll get better and there'll be a transition time in the future But right now I wouldn't class neutron is production. Yeah, correct. So don't don't try to use If you're not running it from the hardware, don't try to use VLANs as a software. It's going to be very slow Just try to gerry toenails. Yes. Anybody trying to just go that route. It will save three six months of research Why doesn't working properly? So it's one of the things so and we talked a little bit about that earlier How do how do you work with the community here in other words? How do you get that message back to the open stack community so that the upstream stuff does get the message? So well, that's that's a challenge right now, right? So it's if you look at what's speaking to Scotty it's not like Linux that you have one person who made the decision So it's many different leaders that actually make the decision of how things should go. So To transfer that knowledge to a community is actually a challenge because we actually Have the knowledge to transfer it but by the time that it go to upstream It takes too long for our engineers to actually transfer that knowledge So we're not doing it the way we would like to do it in the future So hopefully that process actually get better so that we can contribute more to the community Great, and and I'll answer that a bit too at HP one of my jobs is to work with customers and and we have a lot of efforts upstream as well as Inside building our own distribution and one of my main responsibilities is to get these customer requirements and move them back both Upstream and within the product itself. So HP tries to contribute evenly on both of those fronts in many cases It's more important that the fix be made upstream instead of you know forking off any sort of changes to that That's one of the challenge to work in what we chose HP healing opening stack is the ability that you have resources You can talk to you and they can help you with this process So if you have important application for your organization You can transfer the knowledge to them and they can deal with all the politics and make sure that is at the Grow with the community to the new releases of opening stack. They actually is compatible with the application and call you have Great So Devin you have a lot of external I suspect internal clients for your cloud What are some of the biggest requests that you get from your internal clients that the open stack technology is going to allow you to deliver? well, I would say we're really kind of in the same spot as Scotty is with getting developers into the mindset of Building these kind of cloud native cloud ready applications Excuse me So, yeah, we're actually putting on a hackathon on Helion Next month to try and you know Kickstart that a bit And so how are you doing that? How are you organizing that internally? You're looking to get internal people up to speed on the technology, right? So you're sponsoring your own internal hackathon, right and we're gonna bring the hoodies You have the apparel the technology right Yes, you always order the t-shirts first great so Steve high availability and resiliency Has come up a lot and I know open stack had a few challenges in that space But in an area in that area, how have you approached that challenge for HP IT? Well, so there's always two sides of high availability when you when you talk to open stack people It's always about the API is that a highly available API and from an application perspective We care less about the API we're more concerned about whether or not my application stays up, right? So it's been It's been interesting to to go from a traditional IT mentality of I have a hypervisor and a pet that stays up all the time and one one VM runs my application And I have to come up with you know infrastructure solutions to keep that that application online versus the the open-stack way of ephemeral VMs and and applications having to be stateless and having to To be able to tolerate failure In its you know, I think to the point that was made earlier. It's it's very much It's difficult for application developers, especially in an old company to Make that shift from from traditional development to to developing cloud applications So ha definitely there's there's challenges in the in the open-stack in the infrastructure itself and the API layers, etc But also just from the application perspective of how do I deploy an application on this type of infrastructure, right? And Devin mentioned that they're doing a hackathon internally as one of the ways to educate What kinds of things are you doing inside of HP IT to help? You know the organization be able to move in and make best use of this cloud Yeah, and so we've we've had internal hackathons as well. We've had a couple of them now We have we have groups Development groups within the organization that that are much more forward-looking than others and many of those are are driving Driving a lot of that behavior Daniel who's sitting here in the second row is one of our application architects who drives a lot of that behavior as well and it's it's a lot of a Lot of evangelizing and and trying to change behavior Great Devin you I know you you're also working a lot of the past level player level And so what are the challenges in terms of your existing applications and migrating them to are you looking to move to cloud native? Are you are you taking your existing applications and migrating them over and help having them scale better? How are you approaching your existing applications currently? So we're kind of just putting a stake in the sand Stank staking the sand saying net new stuff you need to be cloud ready But for you know our legacy applications. It's probably not worth it to actually go back and You know get them to the point that they are cloud ready So they'll probably just remain on VM. We're based infrastructure. Okay, great. It's got I'll have the same question to you I know you've got different Applications internally and you're you're moving you've got a couple big challenges of moving applications and talk a little bit about that And and similar to to Devin's question, which is what's happening internally from your internal clients? What are they asking? So we'll do the first one first The approach we're taking as Devin mentioned is net new should be cloud ready and able to run in a Public or private or hyper cloud, which means the applications resilient the application is restful. It responds to certain Metrics and monitoring of measurement endpoints. It's redeployable. The other thing we're starting to do our hackathons haven't been around open stack We've been doing container Docker hackathons We're teaching the developers that if you don't embrace the cloud ready thing at least embrace the container thing and then Orchestration can make your stuff cloud ready All right So if you can make your application stateless at start and inject state when you boot them then we can restart them and stop Them without telling you and that one's a little harder to deal with because the developers don't like to debug in a container environment as well So between the hackathons and a big stick. We're trying to encourage people in the right direction We're different. We're not doing this from that We've got management direction that we need the agility and the reprovisioning based on unknown applications Dev mentioned earlier. I don't know what Production or marketing is going to do tomorrow much less in a week or two So if they say I need a hundred VMs to run a website It should really just be an API call in a few minutes worth of provisioning and that's sort of the goal Orlando, I'm gonna pick on you first for this one, which is You guys have been using open stack for a while now. Yes So there's many people here who are either just getting started or looking to get started Give me one of your best practices from what you've learned with open stack as as a solution provider in your case okay, so So there is if you want to become a service provider Within your organization so one of the best practices to do I explain to you about the federation but Going to that party is you want to have for example for block and object storage You want to have one platform to manage you don't want to have to to manage for Networking you want to make sure you you stick to one to one the one that come with neutron whether it's DVR or no and just keep it a standard organization and At the way speaking about docker container and applications there will be a The reason why you want to keep it as a federation on the stable and as simple as possible It's because the public cloud providers that are out there make there like the one we're building in the May 29 Allow you to Rather than burst RAM and CPU and storage to the cloud allow you to take new members to your federation So if you're building your private cloud with the standards, there are opening standard that everyone can use you can actually go to a public cloud provider and Get 20 nodes or 30 nodes or 40 nodes and have full control of the of what these nodes do rather than just getting CPU and storage without having any control So if you have applications like David was saying or a scurry for your developer You can actually decide the SLA you require rather than give it to a third party that may not perform for what you need And we use these best practices because the video game companies we have they have 10,000 users one day with the game because so popular they can have two million users So even if you build a private cloud and you want to have these applications If you don't build it with the best practices of allow you to connect to other public clouds and control them You won't be able to scale a video game for example because it's two million users that require low latency and performance It can be any application you have but if you go to the public cloud per se It may cost you a lot of money because you will have to pay for Flavors and profile that cost you more money 16 CPU and core But if you actually have a federated you can actually take control of those servers And make it like the way you want it to do have four color or any flavor you have so in demo best practices Just keep it as simple as you can and Just stick to open the stag as it is Great. Thanks. It's got any nasty the same question in terms of best practices in terms of your Implementations what you've learned around open stack Probably one of the big ones is that you can't hand whittle your open stack environment So continuous integration continuous deployment is key Everything that runs in the cloud has to be provisionable and recreatable by your configuration management You don't want to deploy VM and then go log into it and run a bunch of updates and whittles and scripts because you'll never recreate that state You want to make sure that stuff happens upstream Make sure that you can recreate the other thing if you haven't If you're not familiar with the concept of a chaos monkey go Google it Chaos monkey is something that Netflix introduced as a way to verify that their applications are resilient And that their ops team is paying attention. They randomly inject fault Open stack randomly injects faults all by itself. You may not One of the things we haven't we haven't done it at scale yet But the intent is as you're doing your CI CD processes and as your test cases are running break stuff Find out how those processes recover find out how your application recovers See if your load balancing works see if your IP failover works on that kind of stuff because it will break in productions You might as well break it and test it and fix it in Dev and test Great. Thanks, Steve. You're gonna go next. I know you break things Best practices from your standpoint So I would say definitely, you know keeping the environment as simple as possible you know when to the Anything that you end up having to do to customize your environment makes it much more difficult to maintain If you have if you have integrations for especially when you're trying to enterprise integration and not just deploying something in a lab It ends up being much more complex and doing whether it's whether it's directories or security, you know policies and processes Finding a way to to do that outside of open stacks so that when open stack changes You can have you can have your your perimeter built around it with with your integration services But but don't Don't rely on things staying the same in open stack and don't and don't rely on You know being able to to upgrade and keep up when you know when your infrastructure expects a certain set of capabilities Okay, great Devon You also been using it for a while. So what about you? What would you give these folks here as a best practice that you've learned? I would say lower level learn Python learn Python logging. I'm thinking do it To be able to be able to get down in the in the in the weeds with the absolute on itself. Okay, great All right Folks out there you can start thinking about your questions I'm gonna be able to take a few questions from you guys as well for these guys It's gonna be called stuff the customer See what we can do there. I do have another question though Devon you are Also working with some moonshot hardware correct Could you tell us a little bit about why that is valuable to you and how you plan to utilize that? Maybe cover what moonshot is very quickly to so moonshot is you know kind of effectively small little low power low power low cooling requirements servers in a box 45 nodes in a box effectively so That's a you know we think it goes hand-in-hand with kind of cloud architecture where you have these horizontal scale-out scenarios Anything you can basically distribute you can distribute across moonshot So regarding the moonshot one of the things that we are saying is for GPU rendering It's just really good because you can use What you're able to do with a Xeon? We push a lot like push a lot of money and a lot of scalability if you're running your cloud With temple with moon, so you will be able to get Better performance with the money invests So you can actually save a lot of money running your cloud for a specific application that require a lot of processing In a con in a very like in a very small you for it like for a you to be able to run it So all the video game companies that I do and transcoding company that I have They are running that kind of hardware to to do GPU rendering or any that we call it or CPU Because it's actually cheaper for them So if you don't know about it, and you're trying to run your cloud just check what application you have You'll be able to figure out that these will actually be cheaper than running what your traditional IT is doing Thanks Steve Duff with HP You we HP also runs a public cloud But I'm asking you from your HP IT side of things How do you view the value of a public cloud and how would you integrate it into your solutions? Do you guys look to use that as part of your your solutions that you're building internally and how? Yeah, well, so we do use our own public cloud extensively today, right? We have a large number of applications that That run in that environment and that we a couple years ago migrated out of we'll say other third-party cloud hosting environments and As we move forward with with With helium open stack. We're we're not wanting to Build an instance of it everywhere in the world where we want a presence And so our intent is to is to run a hybrid solution between either between ourselves Our internal core data centers and in the HP public cloud or the HP enterprise services Virtual private cloud offerings so that we can extend into those data centers But and have the same consistent experience across all the different sets of infrastructure Great. Thank you. All right. I was walking in the halls and I overheard a comment that opens that doesn't work And so yeah, so would you answer that Scotty? It works For certain values of work, right? It depends a lot on your workload on your environment on your skill level where your applications are But for us someone I was talking earlier one of the big drivers for open stack is a consistent set of API's and a consistent tool Set to do automation. My goal is to replace the human middleware component of provisioning of installing a an image of Providing a host I can cut out that whole human middleware server group Interaction I can hand my developers the pointy end of an API and they can spin up a machine They can do their testing they can tear it back down again the and I think it's an overused term It was brought up again this morning developer productivity not having to wait to get your Cassandra cluster even five minutes Makes the developer think they're a lot more productive and therefore they will be so self-serve Open so we've been using it for a couple years now as a self-serve developer platform You can have as many cores bytes and network bits as you want up to some quota limit quotas are important by the way Open stack quotas work We had originally started testing in the public cloud But you get real bills when you use in the public cloud and people don't like seeing hard money They'd rather buy capital as you soft money So it works we're moving pretty rapidly to our first production employment layer this year So we trusted that much Thanks, let's sprung that one on you All right, I will open it right now. Does anybody out here have a question you can Direct it at any one of them or just to the panel itself Is our microphone working? You don't mind going to the mic there we have that working Open stack Did you consider other distributions of open stack and why did you decide for helium? Was that for Orlando or any one of them any one of them. So the question is did you guys consider any other? Distributions of open stack besides helium We did so before there was a helium it was just a glint in Tim's eye We Actually tried doing it ourselves. You can download all the components. It's open source I don't recommend that unless you're a masochist because what's missing in the community open stack is a great collection of tools and an Architecture for those tools interact. It's a platform There's no community installer or upgrade or maintenance or delivery The reason you pick a provider a distribution provider is if you don't want to be in the business of installing and operating the underlying bits of Open stack before helium we were a nebula customer It was a early way to get no labor open stack in my environment It was early for the for the entire marketplace to nebula folded recently Switching the helium gives me some of that same thing. I get I Think HP has seven or eight of the ptls for open stack on staff. They're available I get an engineering group that's unrivaled I get access to people who can help me solve my open stack problems without having to invent that skill set in house Especially for a poc investing in engineering skill set for a poc doesn't make sense. You can rent that Yeah, I would say basically the same answer Looking looking looking at you know running upstream open stack or anything like that You know if you look at say the Walmart's and stuff like that that didn't go with a distribution You know, they have a massive amount of in-house talent to run the open stack We didn't necessarily want to do that great, so so We actually did the due diligence of carefully analyzing what service provide distributors Of opening stuff were available at the time and we realized a lot of them were not ready Well opening stack itself at the time wasn't ready 24 months ago anybody that tried before So at that time we decided to go with a solid company That is a top contributor of the opening stack community that we can speak with and make sure that Whatever need we require they can actually work with us to help us get what we need to get So that's one of the reason why we decided to go with that HP healing of the stack And Steve did you want to address this question? We'll say that with our initial work with open stack two three years ago whenever it was we did look at deploying from You know from from code initially straight on to our red hat platform And that's when we made the decision for our object store to go with Swift stack because they provided that's that the end-end management solution for specifically for Swift now It would be a little bit difficult for me to decide to go to with with any other Solution other than he lay on but but I do have access to you know an incredible organization of engineering and product folks that can That can get things done that I need done and I have given things internally too as some of our other customers have Getting feedback back in so I have their home addresses and Great anyone else would like to ask a question The mic right there would be great and try to speak a little bit loud that Mike's a little soft I will this question is to all of you would like to know what sort of storage you are Using and what sort of thing any tips or any problem that you face when deploying storage in the at a cloud scale Okay, so the question is around storage what you currently use and what kinds of issues you might have run into with that In general or under open stack. Okay, so under open stack since it's helium We're using the VSA Product of virtual storage appliance product that HP provisions on Proline hardware, which is the default cinder and glance storage local disk for ephemeral storage, and we're using SL 4500s full of disks for Swift and Then in success which will be in a couple months, right Desmond We're going to be we'll probably move the the cinder and glance storage to three part So we have that available in the environment Devin you want to tackle that one for you? Yeah, we're we're running actually VSA on top of the SL 4540s as well and Swift on the SL 4540s. We're also looking at PO sinks staff at some point Also on the SL 4540s. You're looking at Ceph as well as Swift at this point. Yeah So we actually have you it depends on your application So I am going to answer what why we choose the one that we chose But I can give you some advice if you if you have object storage for example that you want to have multi data center Then you have only one choice, right? So you have to go with Swift because it's the only one that is there right now, but if you want to have blog a storage and object storage and You want to be able to have a decentralized Then you can go with Ceph now and you can go with preparatory ship a storage solution, but that they're very expensive So you have to give an example what we are able to accomplish with Ceph We're able to get one gigabyte throughput with SAS disk so If you want to do the same with the preparatory ship a solution will cost you a fortune You have to use this as D and probably have multiple tier but you can create the same with Ceph and be able to do it So of course, you're going to have limitation as I say of how much you can put into a set that's why you need to use the federation and If you use I keep saying if you use that federation you they got the good thing that you can use if you don't use that storage The way I see the future is you will be able to resell that a storage after 7 p.m. Where your company doesn't use it or when you're Big you will be able to sell it to for example the future of the helium network One of the things that hope that the clarify the Question Great Steve depth with HP IT Well, and so we are right now for block storage. We are using the the store virtual VSA And we're and we're doing that largely so that we could play on one what we call commodity hardware, right? Which is the deal 380s And that gives us the ability to to run again traditional hardware everything over IP Being able to to scale the block storage on on on that traditional hardware as we go forward We might end up converting to 3 par we are a big fiber channel shop enterprise IT and You know, we're interested in looking at the opportunities with all flash Ddupe and some of the other features of that platform to see What kind of benefits we could get there as well Great. Thank you for your question Anyone else just say it loud not repeat it Okay, so So let me know if I get this right, but the the thing is we talked about scale Yes, and Orlando you specifically stated, you know, if you federated across multiple clouds We can get that scale without having to go over a hundred nodes The question is really is that we are in a transition state on that is an open-stack supposed to be able to handle massively sized clouds and There are challenges with federated clouds. It's just as there are challenges with large-scale clouds So yeah, so to answer that question So of course there's a lot of challenges, but that's opportunity for people in a community to help you solve those challenges So we spent 24 months one of those 24 months We have to spend three four months just to fix billing because you know when you have a federated, but we're a service provider So if you're a user It would be probably your billing to your customer would be easier as if you are a service provider because we have multiple users And we need to allow them to build the internal organization. So Is federation a transition the way I see it with the information that I have I don't see it as a transition I see as a fact that's what I repeated so many time and the reason why I see that is because I'm This is a community that is trying to create a better product and there will be innovation every six months And if you start modifying the code It's just not going to work for you. You won't be able to they say you won't be able to release the package You won't be able to to do it. You have to the only way to grow it is a federated and public providers are actually going to allow you to connect to that Public resources and and have your own infrastructure running on there. Well, it will be There physical equipment, but you will be able to run it like if you were in your private cloud So you don't have polling or orchestration in your private cloud and in your public cloud And which you have a bottleneck because you cannot scale your private cloud except buying more physical servers Okay, and I'll answer from HP's perspective in that We continue in every release to push the the limits on Scale and so yes, our goal is to go well beyond a hundred physical nodes in the clouds But in terms of creating a stable enterprise grade Supportable product. We're kind of where we are today. So I do see it more as transitional We'll break those barriers, but you'll run into new issues once you do that So once you're at a thousand nodes or two thousand nodes or five thousand nodes So you probably have some networking issues to challenge it to you So when you look at the whole of OpenStack That's kind of where HP approaches it in terms of the distributions were created I thought yeah, sorry, I may maybe didn't explain that properly we use set we don't use any HP storage Yeah, so the the question was that in Paris HP got up and talked about how we Provide choice choice and hardware choice and drivers choice and how you choose to put it together And the question was you know, how come everyone's using HP hardware and HP storage? It's good question HP is actually building programs to allow Certification of that additional hardware. Those are being put in place now Yeah, yeah, and we we will continue to add additional hardware. So yeah in the booth We have a bunch of Dell stuff in the rack and some other non-HP hardware Dell and I yeah, so so the short answer is at HP we do want to offer that choice We are building a certification program so that we can work with those hardware vendors so that they can actually Certify their hardware get the little stamp of approval that it certified running the scripts and the tests to make sure that it It's it's supportable So we are moving that direction I would say stop by the booth and ask them specifically on anyone that any specific hardware you're working with I've actually done a little work with Hitachi as well. So we're working with them to certify some other stuff now Okay Am I out of time? Sorry, I'm out of time these guys will stay up here for a little bit if you have a quick question to come on up I want to thank you for your time and please fill out the cards and drop them in the back. Thank you