 All right, good morning So my name is Robert Shrieker I work for SUSE and today I will Talk about the cloud a little bit and then hopefully we will get a good discussion going in the room to see how we as a technical community can get away from the hype and Actually convey to people that are interested in the cloud What this thing really is? Because I think when we talk about cloud in the technical community most of us have a general idea of what people are talking about But once you go outside the technical community to meet a customer or somebody else You will click quickly find I think at least that's what I find That they don't really know what this whole thing is. Of course everybody has heard of Amazon but yeah, well and That is the premise of this talk and again Hopefully we can get a good discussion going the session is being recorded. So if you have anything, please raise your hand and I will Come around with the microphone Okay, so if you believe the hype that we've been in now for what two years maybe three years Three four maybe longer. Yeah anybody else another number five years of cloud hype. Maybe yeah Then the cloud will magically solve all your problems It's just there and every all your data is there and every all your problems will go away Unfortunately as we know That is not the case. So what's going on here? Okay? If you look through the daily press even still today after four or five years of cloud hype When we talk about the kernel for example, right? What you find is there's talk about file systems that we support There's hardware support net working stacks. We even heard that in some keynotes yesterday How BSD supposedly has a better networking stack in Linux, but whatever Right when we talk about the kernel somehow we manage to get to keep people on track and talk about the things that really matter networking file systems, you know Virtualization things like that But when you look at people talking about the cloud, all you hear is the fluffy stuff. Yeah self service. It's always on Magically all your data is there. Oh, yeah, they're these cloud service providers and everything just works So is that just me does that bother anybody else? Well, it bothers me So last year at Linus con so even at technical conferences, I said in the talk of What I call the four horsemen of the a cop apocalypse What was discussed cloud apis and cloud lock in and that may all be well and fine, but interesting love Nobody actually mentioned the underlying image technology, which is really where you get locked in, right? You cannot take an image from cloud a and stick it into cloud beam Unless they happen to run the same infrastructure So you cannot take the Amazon easy to image Stick it into rack space or HP cloud or Dell cloud or whomever else is out there because they all run different Infrastructures, so this is where the real lock in occurs not at the API level Almost everybody a prop provides an easy to compatible API and that's fine And even if they don't it's a number of it's a matter of learning a few commands, right? But the real cloud lock in happens at the image level Nobody appears to be talking about that Unless I'm missing something So I think that we should finally fix this screensaver bug That turns the screensaver on when the machine is in presentation mode But that Ha ha ha Yes, if it were in the cloud, right? so Maybe there's not enough technical stuff to talk about but if you look about it, right when you look at cloud You have to talk about storage. Obviously images have to be stored, right? File systems are important You have to talk about memory management Networking image creation hypervisor support IO optimization virtual IO virtual network stuff So there's plenty of technical stuff to talk about it. It just seems It is either hard to find or it is not there. Oh, that was interesting ringtone Okay We am where's the leader? Today in the private cloud. I claim that to be true and I don't think that many will dispute Way back when when Linux first made an appearance on the scene The free software foundation had a list of killer apps that was widely publicized and they decided These are all the apps that we have to have on Linux to make it successful on the desktop and In other areas Well Where is our list that we have to have to be VMware out of their business? VMware is very expensive and they have a lot of features We are not on par in the open open source world with the features that they have But where's our list, right? Where's our list in the free software community that says This is the stuff we have to have and this is the stuff we need to concentrate on It's not a roadmap. We don't do well with roadmaps. Well, we need targets Here are some targets we could pick up Maybe the cloud community is too fragmented. We have open stack cloud stat eucalyptus open nebula. Those are the most popular cloud frameworks Or maybe we have insufficient communication between the projects, maybe we need we need a more concerted effort at the LF Like a you know cloud working group that creates this list or whatever So you can see I have more questions than answers So what do we need some good old-fashioned flame wars and mutt flinging? Maybe that's what we need, right? Maybe we need to get people riled up and say this is you know, there's a lack here This is where we need to go. We need to light some flames Okay What can we do as tech techies to dampen the hype steer the attention to what matters? and You know what is responsible for pushing the cloud? What is our responsibility for pushing the cloud and That's it. So these are all my questions and things that I have to throw out and from here on It is open mic morning. So hopefully it didn't stay too long at the party or at the off after-party party and with that Who's first yeah All right So I guess the question that comes to mind so you mentioned for example VMware is You know ubiquitous and they have a feature set and quote-unquote We don't have the features VMware has And I think one of the things you mentioned is you know, okay Who's who's who's who's worried about that? Is there any any place where you can? Where someone can point out or can talk about what features VMware has that say open stack doesn't have and you know Even just some kind of like informal like well when I used VMware I was able to do yada yada yada and then when I went to try open stack It wouldn't let me do that or something like that just any kind of information at all you know looking for a starting place I guess right and That is what in my opinion is missing anybody have any answers to that No Okay, so even the companies and I claim us as partially responsible We also have a cloud product even we don't have a list right. I don't think IBM has a list Neither do the redhead guys I think So collectively I think we're falling on our face right we have a serious competitor We have products all you know all of the vendors have products and Still we're missing the target right? Maybe we have reached a point like John had in his presentation yesterday if you saw the weather report There should be taillights because VMware is clearly ahead of us and there should be taillights and there should be a target But yet we're seem to be stomping around in the dark So I don't know What do people think who's who should be stepping up here the guys that make the cloud products Should it be coming from the community? the open stack developers cloud stack Or is everybody just out free floating around? So I guess I kind of like to step back a bit. Sure Where do we think we need to abstract the clouds? Are we are we back to you know? Trying to solve the problem at the OVF level So that you can actually move the images Hold images around onto her into her into a cloud deploy Or do you want to abstract at the API so that you can just move the applications around? I think it would you know at least from my perspective It would be helpful to understand Where you think the problem is it is it at that OVF level so that you can deploy that You know a whole virtual machine image into a cloud and just run the application plus the where or what it was needed Or is it just solely at the business application? Well from my point of view The applications actually don't come really into play an application would be bound to an image that runs on a given OS right and So that image is the workload What do we do with the workload right so you mentioned one thing you can move it around and that I think is where We're potentially still has an advantage with a V motion and other things But we may be behind But I think there are other features that VMware has that you know as far as management is concerned Um That we you know, maybe maybe a functionality is there, but it's not as integrated not as well integrated, right? There's performance monitoring, you know, I mean obviously everybody can launch an image up and down. That's easy, right? Like switch on and off Every interface can do that. We and where can do that the motion I think we're people a bit behind and there are other aspects But I think, you know, it is around the workload and not at the application level And I don't think that the API level is an issue because in the end You know, if somebody decides to go with cloud a either private or public They're gonna learn that API and if they have to move it learning 10 new commands is I don't see that as a big deal, right? collectively we managed to move from, you know CVS and SVN and whatever was popular, you know A few years ago to get and almost every developer and Developers are hard to convince had no problem of learning, you know 20 or 50 new git commands and now that works. And so I think at the customer level we have the same thing Wait, wait, it's getting recorded. You need to speak So so to be clear then what we're defining is the ability to manage Image layer, yeah, including the OS any middleware plus the application. Yeah, that's what I would consider Gordon and I'm going to push back a little bit on that view of cloud because I would argue that is a very traditional server-centric view of How applications and services run that is, you know, I mean even worrying about something like vMotion, for example As a view that you'll be stateful application workloads, etc. You need to protect as opposed to these kind of ephemeral Very fine grained type of workloads that are what the IAS is we have up there really designed to manage Okay, fair Yeah, and I mean the motion mean and and I think You know as far as vMotion and moving images around it is probably also more hype than reality, right? It's one of the features that is on a checklist often and may not be often used while Standing up an image and then letting it run forever more On, you know, a given system until the system falls over and then you bring up another image somewhere else may be more important Question is are we Are we Communicating that message sufficiently So that you know when when you go talk to somebody about cloud You know One of the things that I often hear it's like well vM. Where does this can you do that? that's like Well, do you really use it and Often the answer is no right, but this is the conversation that I don't think is happening because You know people are focused on Other cloud and it will all just work and it will solve your problems instead of being focused on What is it that people really use a me? So I I think that there are certain use cases Motion is important, but I think they really do there are other use cases There we motion like functionality is irrelevant if you have high ability built into the application You know having a sort of a shared nothing model then it doesn't matter It's extremely important to know what the use case is and basically use cases map it to a certain feature Yes today, I mean We agree that vM where has a lot of market and it's just not about the hypervisor Maybe about two things Management application We have any comparable comparable Virtualization application, let's say We have parts of the front right it stops very often with it stopped very often with a K So that is one aspect second aspect typical the M. Bear administrator, I think is Is Has got used to an extremely for his aspect a very nice Typical, you know, gooey licking your eyes, mouth, whatever. We don't have that for For a management or KVM or But now that is missing in our management space And thirdly is the fact that we mware Has basically gone even one step higher in the stack With all its orchestrator Facilities now this orchestrator is extremely we mware specific Our customers don't even notice because it's so used to now writing all these orchestrator scripts that they've deeply Tied into the whole game there and now they come over and want to use Clouds using And some of the final oops all my orchestrator scripts in which I've spent hundreds and thousands of hours We need capabilities there So as to get these just as we have capabilities today to Migrate an image from a VMware file to a QA move Scripts, I mean if I think it's a multi-dimensional problem again, I'm not So I just wanted to point out that at least on the on the storage front We are making progress Addressing some of the future gaps You know on The end of this week will be having the annual LSA and then and one of the Major topics there this year will be copy up load Which goes a long way to addressing both? the in provisioning Feature gap and also the demotion functionality Yeah, so so I think Where we're slowly addressing the problems You know that the at the terminal level Yeah, my question would rather be what are we doing at the middle end? Level to to address the other features right and and that seems Although, I mean obviously everybody's almost everybody's doing open stack The other frameworks that you know still viable and are used. I know that eucalyptus is actually quite popular with you know companies as private clouds and And maybe we are too fragmented or maybe we are missing You know some rallying point hopes at the Linux Foundation where we can all come together and say We need to make these lists we need to have targets something to chase other opinions questions comments It's up for debate whether Linux on the desktop is actually you know successful yet I think it is because I'm running one right now and I've run it for years and it works great for me But the reason to bring this up is I think a lot of what we're discussing is the same as what people used to discuss about getting Linux on the desktop and what kind of thing was the proprietary operating systems still beat Linux at and what needed to happen What really a viable up and so I think rather than than asking the question are we doing enough? I think it should be are we done clearly the answer? No, we still have a way to go But I think we have a really good foundation We have we have competitors to VMware that are clearly viable enough that we're talking about in here They're open source have been done collaboratively. I think those those core things have been done, right? And and now it's just we have to keep doing what we're doing I don't think anybody has has has thought that we're done and has stopped working on it There's lots of companies working on making these things more viable options But I don't think we need to intentionally Bring all these things together. We don't need to say okay open stack cloud stack people of this We need less fragmentation. So we're all going to work together I think the solution is what happened with Linux on the desktop A lot of people came up with different alternatives. There were projects that compete with each other There's still competitors out there, but over time the winners emerged Different solutions became more stable and now most people With a fairly minimal investment of looking at the options and find something that's going to work really well for them And I think that's what needs to happen with cloud is over time Certain products will become more stable. They will become more competitive with VMware And within a few years, I think We will just naturally see some of these these options will be really good competitors to VMware just of their own merit I think in terms of the fragmentation and whether or not we're seeing those pale lights there I think that will not really resolve itself. I think we just need to remember what I've done We need to keep working on it. I think those are very good points But I also think it brings up the an issue of whether we're doing enough in the area of desktop virtualization People may be using Linux individually, but for large-scale and much use you need to virtualize desktop So they can be moved from device to device and I don't think we're doing very much in that area Whereas VM where Citrix and Microsoft are definitely moving into that area. Anything else? Here ladies are not sure the topic is but I just want to share some experience user experience I am I am a unique support person I also have forced to use a Windows desktop and my company like stuff So a lot of things I must use it and I find that all the tools Like you know, Internet access and things I've given is compatible with Windows only and even though China use Linux Personally I find all these challenges Trying to you know seamless interface my work with my personal stuff So that's that's a challenge I have so need to get a better support user seamless user interface Yeah, well, I mean most of the cloud interfaces are web-based so they're really Platform-agnostic, so I I'm not I don't think we have an issue They are other than our web management applications may not be as nice or featureful as the VMware desktop application so I'd like to actually take issue with what you said. I think While what you said is true. It's also unimportant now, right on the client space We don't do so much processing these days All right, the importance of this is that all the processing is moving to the cloud Right, it's moving in the server space. So this is important. We can't leave it to just You know ferment over the next three or four years and on work out You know, he's got the best things who does the right thing So I think the point of this session is actually much more important in terms of you know, what we do You know what how we go about doing that whether you know where you abstract what you try and replicate In terms of the actions that the MAs got the points that you made how you do Injection into binaries to monitor the performance how you do, you know, what what interface So I think it's actually much more important than the client space. So I think it's got a little different. Yeah I mean, I think we are in some respect. I think we're making great progress with those That will eventually be in the cloud the application providers. I think there is progress being made Many of those actually do I think understand What it means when the application runs virtualized Which is you know in the cloud But I'm not sure we're making good progress with those that eventually will run the cloud in the data centers And I think that is getting ground out by After four or five years of hype Still the current trend of well, it will all just work when we we as a technical community We know better, right? It will not just work. There's a lot of work that has to go into that point anything else So part of what I think made Linux work And sort of dominant in a lot of spaces was not necessarily, you know, we talk about the chasing taillights thing where we were supposed like we were all we were doing supposedly was You know, whatever Microsoft and Apple did we would copy it and then that was our route to success I think the actual truth is that a lot of what made Linux work Besides the obvious thing, you know the fact that you know as they were talking about in the keynotes yesterday the The flexibility the agility the freedom aspects of it I think the other thing that made Linux work was the fact that we ended up We didn't just follow the paradigms that you know Apple and Microsoft and Oracle and those people set for us We said our own paradigms and we said hey we can we can do things differently and the way we do it is better So I guess the question I'm asking and I don't know the answer to this is To what extent do we have an opportunity to sort of one up VMware? And the people like VMware and say, you know, okay. Yeah, you don't have I don't know And I'm just making this up. I have no idea. We don't have orchestrator scripts, right? But we can let you write your you know orchestration in You know directly in your favorite language of choice You don't have to use ours a special tool You can just do it right in this in familiar environments that you're already using for all of your other infrastructure It just works the same way. You don't have to learn anything new and it's not proprietary You know, that could be an advantage if it existed Those are the types of things that I think might end up being the answers to how The open cloud beats someone like VMware not in being better than VMware at VMware's game But at being better at being a cloud service in ways that VMware can't match Anything else? Thank you for your attention and Keep thinking about it. Maybe we'll come up with some great solutions