 It is now good afternoon folks welcome back after the the 20-minute break We're just about to start our next session with Pete Bray If this this is your first time in this room today, then welcome to the HP track a couple things I want to share with you about our track today and the presence that we have here in Vancouver for the summit If you haven't checked out our lounge yet, I definitely highly recommend it It's on the west side of the building. It has a great view and There's s'mores and some acoustic music as well as you can get some really nice hoodies and Get some badges Ironed on so different badges such as you know for the different projects There's some different logos and whatnot that they have there so you can customize it Definitely awful also want to recommend that to you RSVP to our party on Tuesday evening so if you haven't done so already the supernatural party that's gonna be a lot of fun and Next little last thing before I hand it over to Pete is that we have lightning talks following this session and Besides having a bunch of great presenters coming up here and giving five minute quick talks about different different ideas or calls to action During that session. We'll also be doing a giveaway for a 10-inch Android tablet So make sure you be here for that if you're not leaving the room You're gonna stay for that session Just raise your hand after afterwards and we'll bring you around a raffle ticket for that so We have amped up open-stack Swift with Pete so thank you Pete Okay, awesome. Thank you Cody and welcome today to our session amping up open-stack Swift So with me here today also, so I'm Pete Brian with the helium open-stack team at HP I'm responsible for developing and marketing cloud storage solutions Based upon helium open-stack and with me here today. I also have Joseph George Joseph. Can you introduce yourself? Well, you don't have the microphone There you need that yes Joseph George director in the HP server group for big data and storage solutions specifically around Apollo servers And I've also been fortunate to serve on the open-stack board in my past as well Excellent. Thanks Joseph So our discussion here today, and I call it a discussion as opposed to I'm not gonna sit here and Joseph's not gonna sit here and go through Slides and tell you about how HP is great Rather, I think we want to get a two-way discussion going here And so I'm gonna start it off by just kind of pulling you and getting a little bit of information from you And coincidentally we want to keep it kind of informal So if you have questions no need to wait till the end jump in let's get a conversation going Let's talk about what you want to talk about here We've got prepared materials and we're gonna go through that but feel free to jump in so the first question I'm gonna ask is how many of you out there in the audience have been working on deploying object storage with your open-stack environment Cool excellent how many are in production Okay Interesting how many of you are using Swift for your open-stack cloud storage interesting, okay Great. Well, that's what we're gonna be talking about today is open-stack Swift and what HP is doing Truly making if you've ever worked with object storage, you know that it can be a challenge to build these systems and particularly to scale them We're gonna talk today about what HP is doing to make that experience easier We have these things at HP that are called cloud optimized solutions. You may have heard of them What we're gonna talk about today is one that's called content depot We also have another cloud optimized solution called helium rack, which is really meant for Rapidly deploying and quickly scaling compute What we're gonna focus on here today is rapidly deploying and quickly state scaling object storage using open-stack Swift together with HP hardware HP services the everything that HP has to offer in this particular space If you do have a question, there's a microphone right there and Joseph will cover this side of the room too So please speak into the microphone so everybody can hear the question loudly and clearly and like I said interrupt at any point So we want to talk today about what we've created is optimized cloud solution called content depot You might say well, what is an optimized cloud solution? That sounds like a marketing term Really what it is fundamentally is it's a combination of hardware software Installation services and even operational knowledge Bundled together in a solution that we can provide for you it includes support It includes even financing if you're looking to deploy a private cloud type of solution like this So it's really an integrated solution meant to be more rapidly deployable and meant to be more easily scalable Some of the key features and as you probably know if you know anything about a open open-stack Swift It's mass. It's built for massive scalability from the ground up We can use this system to provide high availability also with built-in features around replication And of course it also supports the open-stack Swift API It's a very scalable system based on HP hardware and Joseph is going to talk about some aspects of that Particularly some of the new exciting things that we're doing with HP's Apollo server line That are really going to bring even further benefits for massively scalable storage like this But it's also supposed to be simple and if you've ever built one of these systems, you know that in order to achieve scale You got to have simplicity You can be eaten alive by the complexities of developing even just a petabyte size system let alone 40 or 100 petabytes And so that's one of the things that we've done Is we've engineered in some of the simplicity and we continue to hammer away at that How many of you in the audience are in businesses or industries where you're concerned about security data security in particular? Yeah It's a big concern and a lot of people deploying these types of systems are very concerned about both data in flight and data at rest And with this particular solution we can provide Encryption of both at rest and in-flight data and we'll talk a little bit about that a little bit later And then of course it's open. It's based on open-stack standards So let's talk first about one of the major hardware components which are the servers that go into these solutions Joseph's going to talk to us about the new Apollo series the Apollo 4000 series. I mentioned security optimized version of this particular reference architecture which supports both data at rest and data encryption using our ProLiant DL 360 line together with the 6000 storage and a product called an enterprise secure key manager from HP So let's go ahead and Joseph. I'll turn it over to you and you can talk about the Apollo series. Sure. Thanks Pete So When it comes to servers, you know, I haven't been in the open source communities for a long time a lot of times you'll just hear find your favorite general purpose node And Multiply it times. However big you want your your cluster to be right. It's kind of the standard perspective. That's great. Del's got you know HP's got a number of You know great great servers That fit that the DL 380 is a good example most popular server in the world But what we what we've started finding is that as you started digging into these use cases You start finding that general purpose is not always the best fit for specific workloads. Okay? So for example, you know a standard general purpose server will have some compute some storage. It's a great general purpose multi-purpose type type model But what the what the Apollo 4000 family has started doing is Looking at how we actually do storage on a server Better than you know, Jen than what general purpose actually is doing So we if you're familiar with the SL 45 40 is anybody I know this not a hardware show So you may not be familiar, but okay, see a few hands when Fox was Up here on stage and Dreamworks was on stage. They mentioned they're actually using that in their environment Is it just me or is this volume going down? Can you hear me? You can hear me fine. Okay good The SL 45 40 right now is one that's being being used But what we've done is now that we started seeing this particular market, you know customers like you are starting to get How a purpose-built storage platform for you know with with some compute how it's actually making a difference We've actually expanded that to a full portfolio of things right now And so whenever you see an Apollo 4 series server You can you can know that it has to do with data big data Optic stored etc. We'll spend a little bit of time on the Apollo 4200 and the 4510 Which we just announced that are specific to Swift in particular. I won't spend a lot of time on the 4530, but just so you're aware. It's a It's a chassis. It's a for you chassis that's got three individual compute nodes and then 15 drives associated with each one of those So if you anything about Hadoop when data comes into the cluster it Replicates it three times so it works really well in that one chassis and you've got the compute nodes that are there Not gonna spend a lot of time on Hadoop, but let's go ahead and get into what these platforms look like Okay, now When you talk about servers you talk about you write the units It's the the note that the height of the server in the rack This is a brand new product the Apollo 4200 Most of you are probably using to you servers the servers are to to you in height Most of your servers are probably at 15 16 large form factor drives in that environment What we have just done is rolled out this 4200 that actually can hold 28 large form factor drives in a 2u space and Each one of those drives can actually be up to 8 terabytes So it is a form factor that a lot of folks are very very comfortable with It looks a lot like the 380 in terms of its in terms of its Height, but what we've managed to do is we actually added another drive cage to this We if you look at the graphic there The front end of that that server actually slides out and right behind it another drive cage And actually in the back we actually another drive cage as well So you can have 28 large form factor drives in this one dry in this one server Or if you wanted to use small form factor drives, we can use 50 small form factor drives here What we've done here is really taken that storage data use case and really expand it out We understand the community likes this form factor. And so what this does now for Is you can actually have 224 terabytes in that same sort of form factor? You've always known in love you can now do this in a 2u form factor Okay, and again, this is the first first time in the market and with this solution When it comes to 2u servers HP actually has the densest Storage server that is in in the industry All right And finally talk about 4510 and then we'll stop talking about servers If you're familiar with the 4540, this is the next generation of that a couple of things that changed the SL 4540 the previous generation Was a was 4.3 u and I know a lot of people love that 0.3 But we've shaved that off. It's actually now even for you Why that matters is now we can get 10 of these chassis in Iraq now And so with 10 chat 10 rat a 10 test is in one rack These we've also added six extra drives in the old generation. We can only get 60 drives now We're at 68 drives and each one of those can be 8 terabytes each And so if we do a little bit of math 68 drives times 10 you can actually have 5.4 petabytes of capacity in one rack now And what we're seeing is there's there's lots of ways to get this kind of storage Capacity in your in your environment, but now we're able to try and get it as dense as possible a lot of our customers that are Using on-prem solutions or are looking at a hybrid solution where some of their Open-stack environment is is on-premise and some are in the public cloud We're trying to make sure that we're taking as little data center space as possible And this is some of the innovation that's happening there There's a lot when it comes to power savings or it comes to the actual density of the compute And I want to be clear. It's not a j-bot. It actually does have compute capabilities on the front of it And we're working very closely with our helium teams and the open-stack community to To make sure that we're actually trying to drive a more object storage use case for this So as you start seeing more and more iterations of this coming out and this is a theme you'll start seeing from HP particularly in the Apollo line Purpose-built platforms of what you're going to start seeing so this particular platform purpose built for object storage And you know storage on a on a server Any quick questions on that before I hand it over and I went kind of fast on that But I'll hand over to Pete any questions on that Yes The question was are all these discs hot swappable the answer is yes I will say that there are some other players in the market that do something similar to this where Either you they're not hot swappable or you have to take a portion of the drives or a whole entire bay down Not the case here each one of these for both of these models Any drive that needs to be serviced pop them up on them back in no problem Yes, there was one other question Yeah question was do we support both SAS and SATA the answer is yes We support SAS SATA and SSD drives in both of these models, and we also support mixing them if you so choose And again this goes back to there, you know this was that early design argument Why why would anybody ever mix these? But now as you start getting into you know these new workloads you do want some better I you know I owe upfront and certain drives you want to be able to get quick quick in and out Other ones you want to just be cheaper drives to store the data So you can actually use all three types, and you can actually mix and match as needed Any other questions on that yes 4200 so the question was how does it compare to the DL 380? So they're very similar the engineering teams are actually the same teams This particular model was meant specifically for Storage so the DL 380 has a maximum of 15 large form factor drives And again it was designed to be general purpose any sort of use case you've got DL 380 can probably do a pretty good job of it Here we want it we recognize what was happening with object storage We have recognized what's happening with Hadoop, and we realize everybody's trying to get to a smaller footprint So it still has ILO it still has all the ProLiant manageability all those capabilities There are some you know, I believe there are some PCIe Lanes that we took out to make room for that drive cage So there are some features that we defeatured, but for the most part they're pretty close Good question. Any other questions on that that was more hardware questions than I expected to get out of this audience frankly All right, I'll go tell my server friends in Houston about that awesome. All right. Okay. Okay. Awesome. Thanks Joseph really good And I would just want to pick up on a comment that you made about Workloads and how important workloads are You know Joseph here presented a family of servers that you can to a degree You can pick and choose and use what you need and tune it to your specific workload And that's you know part of what we mean when we talk about cloud optimized solution We can certainly if you go at the end of this presentation We'll give you a website where you can go and download white paper and you can read up on our prescribed reference architecture So, you know based upon HP's experience running public cloud with some of the same types of hardware using open stack We've taken that knowledge and we've distilled that into this optimized solution That doesn't mean that's the only way you can build this solution, you know Joseph talked about the 4200 He talked about the 4510 two different options for servers and actually from HP even just alone There's a lot of different options for servers. You can use the DL 380 for instance So there's a lot of different ways you can build this based upon your workload requirements And you know, that's that's a very important point as you design your solution Tuning it The hardware tuning the software to the specific needs of your workload is very very critical So let's talk a little bit about the networking that goes into this solution because again It's a turnkey solution. It includes everything you need to be able to rapidly deploy We prescribe two different types of switches One for the data plane and one for the control plane or the management plane The HP 5900 top of rack switch, which we configure with 10 gig it's it supports up to 40 gig But we configure it for 10 gig and this is for your data plane So this is between your proxies and your your object storage servers and even out to your public client network For the management plane though, we don't need as much bandwidth And so we recommend the the 2920 switch from HP here Which has one gig built in which is just fine for a management plane I mentioned the importance of data security and encryption And we didn't forget about that when we developed this reference architecture It was one of the key things that we understood in the marketplace So we work together with a joseph's team. We architected a solution to address that particular requirement To be able to provide on disk encryption using a special controller from HP the p431 smarter way controller So with this solution and together, you know, one of the biggest challenges in performing this is keeping track of the encryption keys And HP has a product here called enterprise to key cure me secure key manager Which you can use together with this p431 controller To create an end-to-end Encryption solution that you can also use to manage The encryption keys. So it's really a nice solution if you have requirements around data security Any questions either on the networking components or the encryption components? Okay And the nice thing if you've worked with Swift, you know this But the nice thing is it's a very scalable architecture So as I need additional performance Or as I need additional capacity, I can add additional proxy nodes and I can add additional object storage nodes Independently Now there are some rules and if you go read our reference architecture document You'll learn about some of these guidelines. I shouldn't say rules but guidelines for how you scale And one of the key things in an architecture like this is maintaining balance in the system And so you can read about that again in the reference architecture document So all this is great, you know, we've spent some time talking about hardware We've spent some time talking, you know about the various components of the system, but What does it get you? What can you do with it? What we're trying to do now not that we have a solution like this providing object storage is Start to add application support around that You're going to do something with this object storage obviously, right? It may be and we see these as some common use cases There's lots of different use cases for object storage, but we see these In the customers that we work with and talk to we see these as kind of emerging As the main focal point ones today The first area is just a general content repository And some examples here you could be in the media and entertainment industry You could be in life sciences you could be In healthcare financial services doesn't matter You're in a business where you're generating lots and lots of data And I've actually heard some of my customers call this a digital parking lot. I just need a place Relatively inexpensive massively scalable fairly reliable that I can put this data and I can come back and get it later That's a content repository use case The next area And it's interesting if you go out and talk to a lot of different organizations One of the biggest concerns of cios and the it staff is the prevalence of drop box in their organizations It's a real big problem for them And so we work with a lot of customers who are trying to battle this problem and repatriate that data Back into their organizations. This is an area called file sync and share and what we're doing here is we're working with some third party software partners Cetera and gladinat to create solutions to address this particular problem So now not only are we delivering the basic object storage functionality But we're also going to be able to deliver this file sync and share So everything you know about file sync and share drop box You know with access from a mobility devices Collaboration and sharing on a worldwide basis That's what our enterprise file sync and share is and that's what these solutions are cloud storage gateways There's not too many People who are deploying object storage that don't have a need for backwards compatibility with traditional nas Types of systems. It could be nfs. It could be sifts. It could be samba It could be ht. It could be ftp. It could be you know any any number of different legacy protocols And so we're working with some partners here to develop solutions to address this particular requirement the one I have listed here is panzera and finally a very popular use case for object storage right now is Bringing down the cost of cloud backup and archive and more importantly reducing the latency of particularly cloud restore cloud based restores of data And so these are four different solution areas where we've been working to take this content depot cloud optimized solution and really use it for For use cases real-world use cases A little bit more details in terms of the file sync and share use case You know I already talked about providing mobility access Being able to repatriate that data back into the organizations and ensure security but also auditability And then finally enabling team collaboration on a worldwide basis being able to share files and collaborate Not only within organizations, but with external partners too and doing that in a secure way And as I mentioned, we're working with satara and gladinat. Now there are a whole list of actual partners that are working with hp helium on this type of a solution And what you might want to do if you're interested in getting more information about this Go out to the hp website and look for the hp helium ready program It's our program where we're working together with partners like satara and gladinat to certify their software to certify that it actually works With hp helium and there's quite an extensive list already of different software partners that hp has worked with Together with helium open stack. So if you're interested Feel free to go out there and get some more information and certainly feel free to contact myself If you want to get more information I mentioned backup and restore and most modern day Backup software now supports the capability to be able to back up not only to traditional tape based or disc to disc based types of devices Now you can also back up to the cloud directly using object storage containers A feature that hp's data protector added in version 9.01 Is an ability to go directly into a menu and specify an object storage container to back up to And then data protector as it's doing its backups will use that as the target It's really a nice feature You can also use that together in this particular slide shows using it with a product from hp called store wants Store wants is a disc to disc Storage platform meant for A disaster recovery backup types of applications One of the nice things about store wants is it includes integrated De-duplication technology so you can de-dupe your backups and if you know anything about backups De-duplication is one of the key pieces of technology you want in your backup systems I talked about gateways and how important that is to provide backwards compatibility for traditional NAS protocols and We're working here together with satara on some Cloud gateway offerings as well as panzera and in addition to others and we have those again listed on our hp helium ready website Again, it provides a seamless transition for traditional NAS based protocols But there are also additional features that are built in typically To these types of solutions that can allow you to do like tiered storage types of implementations You can even plug into multiple different private or public clouds using these These types of solutions. So it's really a nice It's it's a nice way to provide backwards compatibility, but also give you a lot of flexibility Going forward real quickly I want to talk about a couple of customer use cases that we at hp have been working with On deploying these types of solutions and particularly the content depot solution The first one is an organization called beijing union union read and they are a They're an open source service provider based in china And they chose to go with helium open stack and in particular open stack swift in a very similar configuration to content depot to build Open source storage solutions for their customers And they are looking at aggressively growing their business They want to move beyond kind of their core customer base, which has been in a very Specialized market they want to grow that out into adjacent industries like health care and insurance and other industries and so They chose HP helium Strictly because of that because we were able to partner with them and we were able to help them put in place A solution that was going to grow as they needed to grow And another customer that we've been working with is called okinaw crosshead Based in japan and they also are a service provider servicing smbs throughout japan They have a product called nazda cloud Which is an open stack system for public cloud control and monitoring They as part of the solution offering wanted to build a disaster recovery solution using open stack swift And again, they chose to work with hp because of the maturity of the hp offering And what we could do with them to help them grow their business long term So they deployed a solution Using swift and using a configuration similar to content depot for their object storage solution So it's really exciting to see, you know, these types of real world Use cases come together and people successfully deploying open stack swift using a lot of the things that we talked about here today So I want to open it up. That's actually all the slides I have for today Does anybody have any questions or comments anything you want to talk about regarding swift or hp and our involvement? Yes, and if you can Come up to the microphone, please that would be great And I do have some of our technical experts here today too if we we can go into the technical details if you want to I would like yes to comment that Well, you are working on you are focusing on open stack swift, which is an open source solution And why don't you for example for the file syncing? Why don't you consider an open source solution such as on cloud? Yes Great question. So the question was, you know Why don't we consider open source solutions for some of these files and concern? Yes, we do And I believe we have some of them listed on our hp healing on ready website I think you mentioned own cloud. Yes, that's one of them that we've we've worked with So yes in my organization. We use swift as backend for wrong cloud and Well, it does have some issues, but it kind of works Yes, I think I think with own cloud there were some issues with licensing And that's why we didn't do like an official commercial Productized version of that, but you're absolutely right. You can use it and it does work and people are a lot of Functionality already built in and you don't have to rely on some other company to develop like calendar Yep, makes sense Great feedback Excellent. Any other questions or comments? Okay, well, I thank you very much for your time today and like we said, please go and enjoy the hp area Sounds like there's lots of free stuff And kody Can give you more information