 Good morning everybody. Welcome to day one of the OpenStack Summit. How was the keynote this morning? Okay, I'll take that as a positive Welcome to the Cisco sponsored room. My name is Gary Kovorkian. I'm part of the Cisco events team I'll be your host MC and guide for our sponsored sessions today I'm not gonna take up a lot of time on stage because I know we want to jump right to the substance So just a couple of housekeeping notes. I think everybody probably got their badge scanned when they came in If You should also have gotten a little card for our drawing at the end of the session at the end We're gonna be drawing for a Polaroid snap camera. Okay, great Now that everybody's engaged and involved We'll pick up those cards at the end of the session I'm gonna get going right now with saucing up the stack with Chris and heart and I'm gonna let them come up and do some quick introductions on their own and Away we go Thanks Gary I'm Chris Revere. I'm a cloud solutions architect with Cisco I joined recently via the acquisition of piston cloud computing So spent a bit of time there if people have heard of failure with piston So what are we gonna talk about today? We're gonna talk about saucing up your stack on metapod and I'm curious if this looks familiar to anyone This this is also a metapod. Yeah, it goes over very well in Japan So yeah, you know in a nutshell, what is what is Cisco metapod? It's a you know private cloud based on open stack providing a public cloud experience in your data center delivered as a service You know on your hardware in a nutshell and we're gonna be out there throughout the entire life cycle to make sure that Cloud works correctly for you. So what do I mean by that? We're gonna help design and architect it to make sure it meets your storage needs your compute needs We're gonna do the installation completely remotely that also includes 24 by 7 monitoring With proactive alerts problem mitigation and some of the key differentiators We have are actually platform updates including major revisions of open stack So we provide those with zero downtime as well as proactive capacity planning You know first and foremost, what is it really providing what users and developers want? I have this you know one of my friends is a developer for Congress and I asked them You know hey Joel, how long does it take you to spin up a virtual machine? And he's like well I submit a ticket and then it goes into queue management approves it and then about 42 days later I get an email with my credentials to the VM So what we actually see you know you see this emergence of shadow IT where people now say Oh, I have this cool project gonna whip up my credit card go to a public cloud spin some stuff up So what are the users and developers want they want self-service integration with existing tools Well documented apis and they want speed right speed and agility So We talk about dev ops in action Which can mean a few things to different different people But like you know you take all those logos of those stickers and as we see applications being developed today with more agility What we have is the applications being broken up into smaller components or services You have a continuous integration server That's automatically you know seeing the changes that are being made It's automatically kicking off testing of those instances Reporting back results and then actually when it comes to deploying the application It could just be as simple as you know changing a proxy server Switching something at DNS. And so the point of this is really that all these kind of workflows All these tool sets are a perfect fit to run on top of a cloud platform like like OpenStack Well at the same time what do administrators want right they want to you know govern their users They want to hook into integration with their existing security policies authentication They want high available and they want reporting right control reliability and visibility and metapod is providing all of these components So some of the three major advantages is this performance. This is kind of a noisy neighbor situation But you know if you're running you have control of your own private cloud You don't have the risk of you know sharing a hypervisor with somebody else And someone starts to do some sort of high-performance workload that could adversely affect your your virtual machine And you also get the benefit because it's running You know in your data center on your hardware if you have you know super fast SSDs with you know next generation CPU or memory You're going to get the benefit of that horsepower that you have on premises There's really no lock-in with our service. I mentioned it's it's deliver as a service We have a 4-9 SLA for the entire stack and if you're not happy with the service There's nothing to really stop you from just taking those virtual machines You know out of that cloud and putting it into some other sort of cloud if you'd like and Lastly because it's your data center your own hardware you have control the physical security, right? It's your physical security posture. It's sitting behind your firewall, etc We actually have a number of customers who have deployed Metapod in different countries where there wasn't a public cloud There so they're able to kind of spin up their own private cloud and whatever countries they need to So I usually like to pull like okay, everyone's familiar with open stack But I usually like you know, what's the first word that comes to mind when you think of open stack just anyone complicated Come on keep going. Don't be shy complexity Yeah, all right So I like to say you know that this is what open stack looks like right if you you know deploy it from upstream And the reality is it's it's hard right? It's difficult. It's frustrating It drives a lot of people crazy, and I've you know, I've talked to people where it's you know How long okay? I was playing around with open stack How long did it take you to get up and running four and a half months and Chris by the time I got up and running a new version came out because there's a six month release cycle How do I upgrade to that new version? Oh, I can't find much support I guess I just start from scratch with the new version right so I hear a lot of confirmation of that So that's where you know if you deploy it from upstream lots of times It's not highly available out of the box. It can be very difficult to upgrade And then if you're you know doing it yourself You also probably need a separate test environment that you need to maintain so you can actually test your upgrades there But what we found is really companies want the features of open stack without all this complexity Right you guys want to focus on your business Whatever you're doing at the high higher level the application the agility of your development and not really focusing on this spiderweb mess down here, right? And lastly, you know, we all know there's no 1-800 open stack when it comes to when it comes to getting support So you know we're talking about Open stack it's kind of nice because there's just you know many many options and you know You're all familiar with lots of these this includes some of the newer modules But essentially I look at it as you know the ingredients for making your own cloud So, you know, there's the compute module. Some of these are more mature than others There's the dashboard. There's object storage block storage and they all have you know project names Some of the new ones even have like mascots like this little, you know bear with magic wands or something So you have all these different projects and it's nice like if I look at them as like Lego or building blocks, right? So if I was to give like everyone in this room all these different building blocks and you were to say, okay Let's go. Let's go build that cloud I'm quite sure that there'd be a lot of different clouds that would be built Some people might you know say, hey, we need this specific functionality Someone else says hey, this is in production. I don't want to use something that you know just got delivered a couple of weeks ago so Everyone would kind of build these snowflakes if you will and then you have the risk of you know How many people are actually using these components and how many you know? How are these different versions interacting with each other and So that's where this is one of my favorite favorite features of the open stack website But they announced it in Tokyo right the project navigator So this kind of shows the list of a bunch of the projects with the maturity and we get this a lot when we're talking to you know Prospects people say, you know, I want is a car. It's this messaging service component. Well, the maturity is a one out of eight It's been around for two years and it's used by a whopping 1% of the community If you're building a production cloud, do you really want to you know be part of that 1% using that module, right? There's potentially some risk there Whereas if you look at something like Nova, you know, everyone's using it very mature and it's been around since the dawn of open stack So then we get into metapod. I like to refer to this as kind of a curated open stack And what this is is this is the core components that we're using. We're very You know, like I mentioned, we have a 49 SLA We have a lot of experience running thousands of hypervisors for many clients with these core components So these will you know, you'll notice it doesn't have Some of the absolute latest ones, but we'd cut rather air on you know supporting something that's going to be production and robust So that's kind of where our focus comes from and that's how we can confidently offer a 49 SLA for the entire stack So there's there's lots of options when it comes to okay, you know, maybe you've decided on open stack What are a few of the different routes you can take? Well, we've already discussed, you know a little bit do it yourself Where it's going to require you're not really going to have SLA's it's going to be built on the expertise of your team There's a few kind of famous cases where some large companies built this open stack cloud and then they lost all their engineers They got poached by another company So now they're kind of left. How the heck do I support this cloud? Right? How do I upgrade it? Etc. You're likely going to need people, you know around the clock with a bit of open stack expertise Another option is the open stack distro. So actually as I mentioned earlier, I came from piston cloud computing And it was nice because we could work with all these different components in the open stack ecosystem But one of the challenges that we had sometimes was a customer would have an outage and they weren't exactly sure who to contact first So they would say oh, we think it's something with open stack So they would open up a ticket with us We would spend you know hours looking in the ticket and then our engineers say oh It actually looks like something with the you know software-defined networking vendor So we now have to maybe take that ticket send it off to that vendor they look into it They say oh we think it's actually a misconfiguration misconfiguration on the switch So now you send that ticket back to the customer the customer is like hey I opened this couple of days ago. What's going on? So it can be a little more difficult to actually troubleshoot, you know root cause when an issue occurs And I think that kind of leads us into this other the strength of metapod where it's essentially because it's being run as a Service by Cisco we can support the entire stack with a 49 with a 49 SLA And so essentially the control plane that I'll get into is is Cisco kit and we can offer a 49 SLA on that So what does metapod actually look like? You saw the Pokemon version earlier. This is the Cisco version So essentially I like to refer to this as kind of the the brains of metapod if you will And so this is where we have the 49 SLA. This is to to Cisco as ours To Cisco next to switches everything's in HSRP and when we have three UCS servers Essentially running the core the open stack services and then it's really at the bottom down here that up there is just the control plane Then it's pretty much bring your own server right doesn't matter what vendors out there HP IBM Dell Cisco So you can you know bring your own servers and we have and then you can choose what sort of storage that you like We're very flexible in terms of storage and this goes back to earlier when I was saying We'll help make sure that you have the right storage in your environment so we can support, you know We hook into a wide array of different storage solutions NFS solid fire net app you can use ephemeral or seph storage and the nice thing about this I've mentioned that's really you know production robust is that this configuration here scales upwards of 400 physical servers Essentially just add more switches as you expand so it's you know very robust and a 49 SLA So then kind of we talk about saucing up the stack and I think this is something that We encounter quite a bit and lots of times you you know You show people open stack and especially when people are new to open stack They start asking all these other questions right can open stack do this can open stack do that and the whole idea Is that you know open stack is providing these you know open set of API's that you can leverage up the stack Right, it's not necessarily native open stack functionality So I kind of look at it is how do you want to actually consume it? So we have you know metapod. That's our open stack providing the API is the core compute network and storage And I think what's nice about it is you can have a bunch of different use cases that all coexist in the same environment So for example, I kind of consider it the DIY you have your you know developer Who is just used to native lead developing these you know Cloudy applications if you will who just wants to consume Okay, who just wants to consume the API's directly right and those people are great And that's that's probably one of the most efficient ways is just sitting right on top of those API's spinning up virtual machines Using your existing configuration management tools and pointing it towards pointing it towards these underlying API's But there's other things that can sit on top of that as well such as you know platform as a service when you're looking at things like Pivotal cloud foundry or Apprenda which can actually help enable you to add additional functionality to your applications Whether that be making them more hybrid cloud ready So you can have your application running in a private cloud the majority of the time And if you need to scale you can easily scale out to a public cloud Or whether it's you know picking up existing Java net applications and allowing you to you know more quickly Move those in an open stack in a cloud environment So pivotal cloud foundry is quite nice because you know we've been in cases where you're talking with people and you're talking like this company and they have a developer who's been coding in whatever language for like 25 years and You know you're talking to their manager, and it's like okay. I want to I want to move this application on this Cloudy platform Well, our guys just don't have any expertise coding in the cloud And so that's where I think pivotal has a really nice story where you can essentially take that person They actually go to a Pivotal labs office. They have them set up all around the country They actually joint program paired programming with a you know a pivot is what they call them and they're actually doing paired programming They show up at 830. They got this like gourmet chicken and waffles breakfast They're only allowed to work 40 hours a week And then they kind of rotate with the paired programming so different people and they'll do that for you know four to six weeks And they happen to be developing on this platform cloud foundry Which is a perfect fit to run on top of OpenStack And so now that person goes back into their company and they actually start building out the paired program they start building out that you know cultural shift within their own company and You know spreading it so now they're actually developing more, you know Cloudy apps if you will and now they become more relevant with Developing these applications on a cloud platform instead of the legacy infrastructure. They were using earlier There's a whole bunch of other use cases as well, you know big data There's things like VDI solutions Some of the newer stuff that Cisco has is there's a project called Mantle Which is all based on open source with mesos and kubernetes As well as clicker kind of a hybrid platform that can you know control your applications through a bunch of different sorts of clouds and Actually you talk to you more about that is heart one of our OpenStack evangelists for the pivotal portion So hello there, thank you Chris So hello there my name is hard Hoover. I've been at Cisco for Since October before that I was at rack space. I was doing Operations for some internal services both running on OpenStack and AWS. I'm very happy to be here talking to you about cloud foundry So first Let's talk about the different methods that cloud foundry is delivered to you So cloud foundry is delivered in three different methods So the first way is through their public offering pivotal web services. It runs on AWS It just kind of works as a public cloud function You log in to an app manager. You're able to deploy applications. You don't see any underlying infrastructure There's also on-prem pivotal cloud foundry the middle one which we'll talk about today And finally a cloud foundry open-source version similar to the open-stack model where there's an open-stack DIY version as well as a productized on-prem No version from a vendor some components of pivotal cloud foundry include routing so there is a router that will Set up routes of your application to specific containers underneath There is some service discovery in there for keeping track of where all these services are running some scheduling to keep track of containers in the cluster There's also some logging and metrics available. It'll stream logs directly to you on the on the right side, you'll see services there those are Not necessarily part of cloud foundry, but exposed in cloud foundry. So those can be external services like for example If you're running a database service on the side or have access to a public database service You can expose that internal to cloud foundry same with big data or object storage You can expose those as service brokers in cloud foundry Cloud foundry also runs on multiple underlying clouds including Cisco metapod or AWS or DOI open-stack or VMware So let's look and see how Cisco Metapod and pivotal cloud foundry work together So I'm going to switch off the slides. Oh my gosh So the first thing I wanted to talk about is the ops manager. This is what your ops team will see when they're working with cloud foundry Let me screen this I won't see that. Okay a minute a little bit so Here we see cloud foundry is talking to an open-stack cloud, which would be Cisco metapod It also has a service exposed. So this is an operations team for pivotal cloud foundry Making a reddest service available to applications for developers and then this is the pivotal elastic runtime Stuff so this is all the guts of Cloud foundry. Let's give that a second to load on our super awesome Wi-Fi So you'll see all the components of cloud foundry here and they all have handy-dandy IP addresses. I'm going to zoom in a little bit more so y'all can see that It'll show you some stats on what's going on with all of those services. Not much. It's just running for us So this is tied directly to instances running on metapod So here you can see all the VMs with all these little IPs here corresponding directly to these Services in cloud foundry. So here's an auth service HAProxy load balancer, mysql, etcd and console, service discovery things So those are all tied directly to VMs and metapod and there are a lot of them Here's our IP for ops manager So that's what an ops person sees With regards to Pivotal cloud foundry For the developer they're going to log into an apps manager so When you log into an apps manager in Pivotal cloud foundry, you'll have access depending on who you are to different spaces or different organizations excuse me So we're in the metapod space or I'm sorry metapod org and under that we have access to several spaces So for example, we can break this out into development production or QA or these can be broken out into different teams within a single organization And so I want to show you the sample app that we have running here so this is running on Pivotal web services so the public version as well as our private metapod version This is running a couple of containers. We're going to add some add some Containers here. I just want to show this app running here. So our our metapod logos here represent the instances and we're adding Containers to those instances So here you can see in the app view You can see the different services that are exposed for this for this app. I'll zoom in again so you can see it better You can also see environment variables exposed for the service So we can have these when we start the service. We can pass in some environment variables very 12-factor And it has a route to our to our application and then streaming logs, right? So error. Oh well You can see my requests here So I'm actually going to scale this up. Hopefully the demo gods don't curse me sweet So cloud foundry went ahead and automatically scaled that up for me and then I can start adding services so again, all of this is running directly on Cisco metapod and Then I get a log of events that I scaled the instances to six You can also instead of scaling horizontally. You can also scale vertically by adding more memory or disk scale back down to four So that's all as far as a cloud foundry demo Thank you for coming Chris and I will take questions now Be sure to stop by our booth C11 for other giveaways We'll do the camera. I guess in a minute or Gary's walking up here. Maybe Q&A first, please approach the mic if you're gonna ask a question Yeah, please just Yeah, we have to be able to hear you Is it possible to run just custom Community version of the CF instead of the PCF Metaphone so the question the question was just to make sure I understand that question before I answer even though I just answered Is it okay to run the open source cloud foundry on on open our metapod or open stack? Yes, absolutely as long as you can validate your Install of open stack against cloud foundry should be fine and there's steps in the documentation on how to do that Are you currently supporting the cloud foundry Diego or just the DEA? So Cisco itself Doesn't support cloud foundry. I mean You and the runtime environment for the droplets within the DEA Have you do you only support the DEA or have you started supporting Diego in place of DEA? Yeah, Diego Diego as well. Do you support both? Yeah, yeah, okay, that's what's trying to find out if you're supporting most current version with the most current version is what we support Want to hear support I think I was just wondering how do people handle Networking for this is it usually all done in one big flat layer to network or is it segmented out? So that's a great question. There's there's a few different options in terms of how we handle networking Our installs are based on neutron For the most part there's a couple different options One of those is Cisco can completely manage the entire network meaning the switches the routers and you essentially can create whatever sort of Virtual networks you like and attach floating IPs to get further connectivity so we can have you know list actually pull the slide So actually we can have lists of you know different IP pools So you might have one that's a range a corporate IP range another one could be public-facing and addresses We have a few other models where a customer can control the layer three environment So customers responsible for the routes the routers and then you can essentially as you provision VMs You can put them directly on your corporate network if you'd like But for the most part all the routing that takes place is off-loaded from the software into the ASRs Did you want the architecture side? Yeah, that's fine. Any other questions? I say you're in integrated ironica. I don't know whether you implement Talent talent and network as a nation Did you get my dear? Do we support multiple tenants in metapod? I know Ironica did don't SNAT network What are all the talent show network in community in Washington? So I don't know this cause why the Cisco implement network as a nation So so what you do tip traditionally in our our neutron implementation is for each tenant You spin up a router and a private network for that tenant So it is completely isolated unless you create another router and tie together multiple networks But by default everything is completely isolated I think we're over on time Or we're okay You're coming with the with the hook Any last questions If not we can hang out Or you know the our Cisco booth C11 opens for business tomorrow morning at 1045 Well, actually, hope we all stop by tonight at during the booth crawl from six to seven thirty We've got some goodies going on there, too And with that said We're gonna I know the cards are already starting to be collected. So I'm gonna go out every now now everybody wants a pen. Okay Actually if you guys can just pass them out to the aisles that'll make everything real simple for us We're doing we're doing we're doing more more more Okay, okay as they leave your odds go up so I'll let you when I actually gonna let one of you guys pull the card so Okay, okay, do we have everybody's card? Nope. Oh Gotta hold this in a whole half of the room here. Is that it everybody in? Oh There's always one All right, here we go where's the camera and heart is gonna pick the lucky winner mix them up The winner is Donna de Capic Donna de Capic. I'm probably saying it wrong D. E. C. A. P. Okay. Yay Thank you, everybody. Yep. Thank you guys for coming again come see us tonight during the booth crawl come see us starting And starting tomorrow morning at the booth C11 we'll see you there we've got another session coming up on ACI in about 10 minutes