 Thanks, those are awesome Good morning Jim's gonna come in a second. Jonathan gave away my surprise, so I won't build to it as much anymore It's really awesome to be back in Portland For those of you who've been with this project since the beginning, you know, this is the city that we started in It's actually the building that we started in we had a little smaller scale We certainly didn't have all the IT operations that are going on behind the scenes now But it's really fun to be back here So I look really smart for tiling my title the year of the user I'd actually didn't coordinate with this. I didn't know we were gonna have users talking today But I did this because I actually think it's important that we start talking about the user And in fact, I'm gonna do something a little different with my keynote I actually want to hand my keynote over to HubSpot who's a real company Using OpenStack to solve real business problems. So let's just keep going with this theme before I do that I want to give you kind of a quick update on what's been going on with us in the community and talk a little bit about how We actually got to this point So in 2010 we had a very basic but important task to do That was we needed to actually build OpenStack I don't think it's a secret at this point three years on that there wasn't a whole lot to begin with We had a promise we had to do a lot of work building this So we took some smart developers out of rack space NASA and some other enlightened organizations that believed in this cause and said Let's get together work designing and building. What is what we know as OpenStack today? 2011-2012 a funny thing started happening User started showing up to this development conference People who were much more interested in running or operating OpenStack than actually building it And so users and developers had to learn how to start working together to build this project And that is one of the major defining characteristics and unique aspects of OpenStack I think in the community today. I think this is the year that the user really does take over So we have almost 3,000 people here this week. We just saw that there's their 500 developers that contributed to this release I think it's safe to say looking at those numbers. Most of you didn't contribute code Most of you are here because you're either running OpenStack today You're helping someone else run OpenStack or you're here to learn about it So I think this is the year in which we really start to see the users define where OpenStack is going and what its future is So that means we're at a real tipping point The people who are running OpenStack today, they're not zealots about OpenStack for its for its own sake They really want to use it to do something they want to put it to work on Applications or any sort of business problems that they may have if you were with us in San Diego We had Troy Toman our head of engineering Give an update on how we're actually using up OpenStack and the performance characteristics We saw in our public cloud so I want to give you an update on those stats today But highlight what they really mean OpenStack is being used and it works. So let's let's look at some of those stats We launched Nova OpenStack compute in our public cloud in August since then we've had over 549 million API requests Pretty good usage, right? It's had 99.94% availability here in that time Pretty reliable and as we've been working with you and the community to build Grizzly We've been deploying it near real time as well We wouldn't do this in an OpenStack based cloud that's running real customer workloads unless it actually worked So in short, I think what you guys are doing is working well. We're very proud to be a part of that community effort So some thoughts on how many people are using this we released Version 3 of our private cloud software about six weeks ago It's free and open source and we've had almost 8,500 downloads of that in that period of time From every customer segment and region you can imagine and a lot of folks who are deploying this on hundreds of nodes That's proof that the world wants OpenStack and is ready for it today If you saw yesterday, we made an announcement About the Rackspace global cloud network really happy about this our intentions to help service providers Who not only want to deploy OpenStack based public clouds to be able to do so but also to help operate them Which is a key component of getting clouds up and running We also want to make sure that those clouds are federated together So that users on those clouds can be assured of a consistent and integrated experience around the world And so we'll be doing a lot of work on that this year People usually don't like to show slides to go down into the right in this case. I'm really proud of this So Rackspace's relative code contributions to OpenStack have decreased. It's a great thing The reason as the community has stood up our relative contributions have gone down You guys are simply developing code too fast and we can't keep up Doesn't mean we're not developing a lot. It doesn't mean our overall level isn't increasing It actually is we're adding developers to the project every day, but you guys are simply growing too fast The fact is there are over 45 companies that contributed to Grizzly Which means that no one company no one organization no one person actually owns the direction of the project We all do it's a true community effort And so one of the things I want us all to keep in mind this week is we're all talking about the different great things We're doing as vendors and users in the community It's to remember that this really is not about Any one company or organization winning in OpenStack This week is all about making sure that we're all doing the right things To ensure that OpenStack actually wins and is successful for users in the cloud market So we have almost 200 rackers that are here this week There are crossing every span you can imagine we have obviously developers We have people who are involved in our qa in testing. We have lots of our operators We have marketing people. We have community people If we can be of help to you and engage with you, please let us know We're here to help build something great with you this week So that my surprise is gone I want to introduce HubSpot So if you're not familiar with HubSpot, they build marketing software That enables their customers to convert leads Into customers for themselves They serve over 8,000 customers worldwide I want to do a video before I have Jim come out to do a quick introduction to what they do and how they're working with OpenStack today So Here we go HubSpot we like to think is the only marketing software that you'll ever need We have over 8,500 customers We focus from SMB through the enterprise and we do everything from Content generation capture social media lead nurturing email marketing all the way to marketing analytics We think we make marketing that people love So we've been a racked space customer since 2007 early on we decided we do not want to be in the Colo business or the racking hardware business So we didn't have the proverbial server under a desk We had a server in texas our mantra is we shipped a production about 75 times a day about 2000 times a week for qa We not only deal with a continuous build and continuous deploy We deal with continuous provision that gets extremely demanding on a team like ours where we have to make sure that infrastructure is ready And able to take those updates at any point in time our cloud needs to support all those types of products that can range from Hadoop all the way to mysql to jango stack You name it for us a product idea might be a pm who has some coding capabilities Who wants to put out a prototype and actually get internal feedback or get a select customer group of feedback And the last thing we ever want to do is tell them they have to wait for it Or they have to wait for us to get a server so from idea to production could be 30 minutes So we've also grown up in large public cloud spaces So as we apply that to rack space and to the future of our open stack implementation there We expect that same provisioning capability. So we will not wait for 30 days. We wait for 30 minutes And that's the way we roll. So we're big believers in open source at large. We contribute we consume 95 of our product has been built on open source But it's always been at the application and library layer up This was the first opportunity that we believed Could handle our volume really in our future growth and scale that we could actually have an open source VM hypervisor slash operating system Which would give us all the benefits of what the large public cloud providers have As well as controlling qos and scale that we would like to see in more of a private Managed environment at rack space. So the the kind of combination of that burstability Plus the control visibility and qos that we can get is really why we chose to go with open stack Open stack allows us to build things that we may not otherwise be able to build in a public cloud We have full control over things that we might find and be able to change Things with the stack itself and push those changes back up to the community We believe that open stack with the community that's being built around it will be over time the fastest way to those paths where It's an api driven open standard infrastructure versus a hardware closed based infrastructure That's one of our pure tenants about why open stack is the choice Rack space has always been great. It's great to see that they embrace open source so much The fact that rack space cloud is running on Open stack is fantastic And the fact that we can build hyper clouds that are based upon a single model is even better for all of us Okay, and with that I would like to ask Jim O'Neill See how it's about to come up and tell us more. Thank you. Thank you All right So how we doing We all doing well Levels are coming up. You guys are recovering from the long trips and the plane rides and all that kind of fun stuff Okay so Let's get our little power point up here So today i'm super excited And if you listen to what jim just talked about in jonathan before him, it's amazing to see How far open stack has come in a few short years? And with that i'm actually going to tell you our history about open stack Which is actually more than just a casual relationship. It's actually a bit Of a love story And with any good love story It actually began with a meeting of chance And so in 2006 two gentlemen brian halligan and dharmesh Shah co-founders Were both serial entrepreneurs and they were at mit and they were studying their MBAs And they were realizing the way that we all Learn shop and buy online was fundamentally changing These changes in consumer habits were driving us from cold calling email direct mail and interruption based To what i think we all use and love today, which is social media blogging and search engines So let me ask all you Are consumers in control? I think consumers are now in control How many of you guys wake up every morning? Excited excited to check your inbox. I know i'm not So then why do marketers spend all their time trying to shove your inbox full of email that you don't care about And the reality and this is what brian and dharmesh realized Was the marketing playbook was just broken So let's look at some stats. Okay 86 percent of us skip tv ads and i honestly think if it wasn't for the super bowl those numbers would be way higher We all have dvrs 91 percent 91 percent of us Unsubscribe from emails that we didn't ask to receive and worse most of us probably marked those as spam 44 percent of all lonely direct mail gets put into the recycling bin And then the last stat is over 200 million people 200 million people are on the do not call list and the last i checked it was only about 300 million of us americans So how's that for an unsuccessful marketing program? So brian and dharmesh back in 2006 Realized the marketing playbook was broken and they not only set out to fix it But they actually set out to make marketing something That people could love and so we did that and we started off hub spot. We've been very fortunate But it wasn't long after that And we had that love We realized that we needed to also make a product that people loved To discreet marketing that people loved wasn't enough And we were actually having a great great time And with that it allowed us to create some great growth So if you look at these are real numbers we're a very transparent company. That's our revenue growth We're enjoying this But there was something still wrong when we started the company and you guys probably saw it in the video a minute ago We did start off in a managed facility and it was a great managed facility. I'm a big fan of rack space But our growth was so fast that it couldn't keep up with us Racking proverbial servers via our partner Or without us front loading a whole bunch of costs and able to do that I'm talking dozens and dozens of servers every few weeks And we were looking at this rocket ship and we realized that something was wrong Along came the public cloud Okay, you all know which one i'm talking about you all do She was fast. She was beautiful. She was unbridled And this let us do things we could never do We had fast We had flexibility and we had scale That scale is what we need And with that we could actually create something That we could love and this was great But with any good love story there's always a little catch Zombie servers For those that don't know what a zombie server is their servers That lose their mind. They lose their storage. They lose their network. They become lifeless But they actually consume resources those resources can be dollars those resources can be your time Ensure here and there one or two. It's not a big deal. But we know at scale zombies multiply and one or two Can become a few And at hubspot scale are running thousands of servers a few can become an army Now that's disruptive and it's really hard So with that We had to figure out operational costs And in our world we had about a two percent monthly instance failure rate zombie servers Okay, you can kind of deal with that but at scale again at thousands of servers. That's get pretty difficult At the same time We have a 20 percent quarter over quarter Instance turnover rate. What does that mean? Well, we're growing so fast And we're having to spin up so much capacity that we may need to scale up Or we may need to scale out our infrastructure And in the public cloud that's hard Because you don't necessarily have all those options to do that So what we really have to do then is to take all of that and do it ourselves At the same time that tugs exactly against our reliability So this is our target uptime four nines and I hear telecom the guys that went before it's amazing when people talk about five Nines will be very happy when we get to four nines But to get to four nines You really need to have a lot Of control And it's hard hard hard hard to have control when you have this kind of chaos So we look to the legends the google's the netflix's the facebook the etsy's of the world and they all told us You have to automate all the things. Can you guys say automate all the things you guys are in the cloud? Come on automate all the things. Okay. Thank you automate all the things And when I really say automate all the things, I mean all the things we had to automate everything Now there's awesome companies out there. I bet your right scales in the crowd. We're huge fans There's hiroku who's in the public cloud But when you guys saw that video you heard us say that we shipped to production 75 times a day We've been doing that for over three years. We've never changed the cadence on that It's a stat that i'm super proud of it's actually kind of crazy a lot of people tell me we're crazy But the automation of all the things made that possible and that was great So with that we could love a little bit more. We could love our cloud a little bit more But seeing how we have this story of zombies and zombies can be very pesky Especially especially in mass And every once in a while those zombies would get really really upset And what would they do? They would actually create an apocalypse Now we've all lived through apocalypses. We shouldn't but we do And they're painful and they're disruptive And many times when we looked at how do we survive these apocalypses the answers are Move to another place in the country that's safe from zombies. Well, that's really Time-consuming and expensive especially for a startup even at scale And what we really figured out was that we needed that more control that we talked about And in these cases we often have to wait for kind of central command to come in And restore civil order And that was hard So we would looked around and we would say what can we do? But we really felt like our hands were tied They were tied to this cloud model that we couldn't control And that got very frustrating So then along came OpenStack And a little history of OpenStack with HubSpot is every year since we started the company And I've been there since we started it We went out to market and we would look at the public cloud providers the private cloud providers The colos and all those kind of folks in between And in 2011 I got very lucky very lucky I'll say that one of our earliest it guys came to me and said you should go to the Diablo design summit It's down the street in Boston. I'm from Boston And thankfully I did I went to that summit And it was awesome I met people who are like minded to us and who shared the exact same beliefs of open source That we did when you guys saw that video and I say 95 of our application stack is open source I want the whole stack. I don't want just the app stack. I want our infrastructure stack as well And so we came out of that conference Extremely excited so excited that we decided we were going to place a bet So within one month of coming back from that Diablo design summit again, we're a late stage startup We put down a whopping nine servers nine nodes now. That's not huge in scale But it had the promise we saw computing capacity We saw that we could take Workloads that we run in our public cloud and we could apply that there So nine months later we decide we're going to double down We go with a whopping nine nodes with Essex And it's starting to get a little interesting because we're beyond just compute We're now into storage and that was a huge need for us Just as a quick sound bite we run about a petabyte of storage It's it's not the big big data, but it's getting there So now fast forward a little bit later six months after Essex with the launch of Folsom We place all cards on the table 166 nodes and that may not sound like a tremendous amount If you look at the number of cores and the amount of ram, it's 2,000 cores. It's about 20 teas of ram It's two and a half petabytes of storage. That's actually larger than our entire implementation Of our public cloud So that's pretty big What's even more exciting it actually doesn't stop there with Folsom With grizzly, which is honestly just two weeks old and we've all just heard about grizzly We're taking that one step further and this is all live We talked about the public cloud Okay, we're going to add in our private cloud for hybrid. Okay We're now extending that with grizzly to full bare metal So when we think about it, we have the exact same image Running in the cloud of our choice that could even be a colo if we chose to And really the beauty of that and what that gives us Is image parity. So I just want people to pause and think about that How many people have been able or been wanting to take that one single os image And run it in any data center including a public cloud And the beauty of that is our application doesn't know where it runs anymore And our application doesn't need to care where it runs anymore and we find that really game changing So with that we can have public private and bare metal all in one That gives us efficiency. So I hadn't talked yet about the efficiency word So we've taken this now single open stack image and we've picked it up from the public cloud We've moved it into our rack space powered private cloud And we're seeing the same exact workloads. These are Hadoop H base So even non-traditional from the VM standpoint And we've seen a whopping four times increase In processing of that exact same workload. So we're really really excited about that We actually think this is just the beginning We haven't yet had time to tune and move all the knobs and the levers because folks that deal with Hadoop and H base No, there's some intricacies with that How's that for control People excited I'm excited. Come on. Be excited. Be excited. Okay So with that story We're actually going to take a pause and mark from con cast and I stole my thumb thunder because You know, we are going to do a live demo And the first rule of live demos I'm going to walk over here for anyone that's ever done a live demo is you never do live demos But I'm going to do a live demo And so let me talk a little bit before I do it about the live demo challenge So when the rack space gang came to me about a month ago And they said Jim, I'd like you to do the keynote. You're up to doing it And I said kind of like Jim Curry. I really I'm not a big public speaker I really don't want to do this, but I'll do it because I feel that strongly about open stack But what I said to my gang who run our DevOps team is can we not just do the standard keynote? Can we actually demo All of the love and magic that we've got in our public cloud And actually make that work in the private cloud in this new open stack private cloud And I threw the gauntlet down and I said I don't want to demo I want the real deal So here's the caveat only 10% of what I'm going to show you today Was tweaked and I'll be honest. I'll be fully transparent about what we tweaked for the demo today Everything else is 100 live So I'm going to ask for one big thing this demo works and I have 100 confidence It's going to work. This is in production by the way real production You give our DevOps team couple in the room Craig Tracy You saw in the video give them a little love when I'm all done with this You've got to give them a little applause Okay So with that I'm going to take an application Just any app our login app So for HubSpot, we have tens of thousands of marketers creating love every day for their customers And I'm actually going to take that app and we're going to make a change and add more open stack capacity to that app today That app also runs in the public cloud So I'm crazy enough that I'm going to take that public cloud I'm going to sprinkle in a little bit of private cloud powered by open stack I'm going to do that in under five minutes. So we are going to go and I'll do this and you guys can all time me We are going to go Take a image from glance on open stack Get it on an open stack nova compute node We are going to provision capacity. We are going to network We are going to do all that fun stuff and I'll go into the details in under that five minutes Now I don't know about for you, but for me That would have historically taken Days, weeks, hours, and maybe even longer in a traditional it environment. So I think this is a big deal So let me introduce you to rainy Rainy is a good friend of mine. She's the actual tool that I'm going to use today to show you this Now rainy comes from the name of raining servers because we were raining so many servers all the time. Okay That's actually a lie It came from raining cash We were spending so much money spinning up servers so fast We've long since fixed it. We've learned the lesson if you guys know the public cloud, you know to turn the lights off That's a very common expression nowadays. We now turn the lights off, but I'm going to introduce you to rainy So rainy is going to do for us is she's going to provision Hardware she's going to install software. She's going to configure that software We use zookeeper and a bunch of configuration tools and we're going to network that software That is going to be live in our production environment within five minutes Now we couldn't do this I'm giving two shout outs and I normally don't do shout outs But I'm give two to ops code and to pop it because anything that you guys see around these automation frameworks is not possible without those guys So give those guys Thank you That is not something hub spot done. We just take those ingredients and put them together All right, so let's do a real real demo So if we can switch over to this computer We live good you guys can see chrome. Okay So we are going to go to rainy and let me shut down my instant messages. So no one pops up now Rainy looks a little sad lonely, but we're going to get into it and a couple interesting things We call it condensing and what condensing is squeezing the clouds We're going to squeeze these clouds into actual computing power and you can see I can do open stack Or I could choose a public cloud And some quick things as you can see a host name This is live don't kill me And don't get me into host names. We treat servers as dispensable. So we use mountain names. You guys may use something else And this is a list of every single one of the hub spot projects. It's 100 automated We have this massively fine grained cloud infrastructure. So I'm going to do the demo I'm crazy enough. We're going to do production Hopefully anyone that knows clouds knows that you tend to choose an instant size now the beauty with open stack. I'll choose a large is If I chose a Hadoop Or h-based type infrastructure node. Well, that's going to go on bare metal So from the end user a developer an operations engineer Again, they don't know and they don't care For today's demo, I will do tomcat and I'm going to condense Aka squeeze the server So now we're going to see What we call the pizza tracker. So we love pizza make hopefully you guys all know domino's pizza It was like this genius thing where they made the pizza tracker Which tracks from the time I place the order for a pizza or a server All the way through to that yummy new server that pizza is ready Now as this goes through jordan, who is a real dev ops engineer at hub spot He's not really behind the scenes making a fresh pizza with organic ingredients for us He's actually taking an image from glance Getting that image onto an open stack compute node and starting to provision that capacity Right after that chow who's another dev ops engineer is taking the wisdom of the puppets and he's getting all of the puppets going So literally in that 30 seconds glance Nova compute have worked and we now have the start of a virtual server Once that virtual server is there and it's on the network. That's when all of our lovely puppetization comes in And while that goes Jim curry had asked me a little bit to talk a little bit more in detail Because this does take about three and a half minutes not too bad for a whole server online in production To talk a little bit more about our open stack implementation So when I talk about those 166 nodes, um, we break that up between bare metal We have about a hundred ish nodes that are bare metal running hadoop So we think of data nodes pure bare metal image parity open stack awesomeness We take our counterparts in the hadoop hbase world Name nodes things that are more sensitive to failure and we actually run that on open stack So we have a hybrid hybrid cloud if you actually think about that Then we take our web and our application servers and our restful services and we run that on a pure open stack environment Say the rest of that 166 nodes We went all in we talked about glance a little bit and imaging We did tend to get networks. We do not want and especially if you guys hear us turning servers so fast We don't want imaging taking us a long time. So that's why this is all possible A couple other interesting things is when we talk about what we use with open stack. I mentioned glance I mentioned nova compute. We also are using block storage and normal storage So cinder and we are an early customer using a net app device for massively sized block storage So we've got multi hundred teas sitting behind that nova compute infrastructure Also going on as this magically goes through this is we've got LDAP with keystone right because we have to authenticate all the nodes to get into this We've got dns going on So there is a whole bunch of magic that's happening behind that So I was taking some time on the Puppet wisdom going on which is surprising. Usually we have a special guest appearance by now. So we'll we'll give it another minute Some other things that I think is as important as we look at what we've done with open stack is the contributions around the ecosystem So I think a lot of people look at open stack is just the computing power and the images themselves I think this gets back to the whole room about how fast the environment has actually moved So let me just kind of be impatient and let me try to refresh our pizza tracker and craig if you're in the audience Hopefully you're out there figuring out what's going on So I don't have to sit up here and talk all the time And this should have been done by now because my my my timer is going so where is craig? What's going on? You're giving me the look Everyone's here sitting this page. Okay. Is it just not refreshing? Okay, so let's talk about this page while craig patiently waits right there We purposely changed the url So if you see open stack rainy dot hub spot dot com that's purposely been changed to protect the innocent We knew a lot of you folks in the audience would be hitting that url What's going to be cool is once this finishes condensing and did anyone else condenser? Was it just me condensing? Once we do finish condensing we're actually going to show you our login application And that login application when you look at the actual destination you will see that is not in the public cloud It's actually in the private cloud Let's do a little refresh You're gonna make me start dancing aren't you? Boo, okay, so We'll go through a couple more quick things What i'm going to do Is i'm actually going to open up our login app Okay So this is going to be the app that that node gets popped into and I guess because everyone's hitting it I think we need more capacity Um Totally embarrassing see that's why we're not supposed to do live demos We've practiced this we've practiced this like a million times So We'll see what's going on So anyway, what I was going to do and then I'll I'll get to the punch line in a minute What I was going to do is go to our login application Our real login application you will see For hub spots point of view Um We celebrate customers. We celebrate people behind us. This would be what you normally would see And I was going to actually do something really cool and live from the open stack keynote In the past couple of days So we're going to punt on the demo there craig Punting on the demo ah the rule of live demos Okay, so with that production crew if we could split back to the power point So my whole punch line Was how is that for control And you guys are in control Because you could have been able to do that So I apologize live demos everyone So with that You should be able to love your clouds even more because remember you've got public cloud You've got private cloud and you've got bare metal. That is a hybrid hybrid cloud So with that we not only love our cloud. We love open stack We're admonishing and abolishing our faith to that today And I wanted to say thank you check all this stuff out on dev dot hub spot dot com We're going to be blogging about it. We're going to be open sourcing lots of the tools that we've added around this ecosystem So feel free to check out dev dot hub spot dot com. Thank you