 So, you just heard Yahoo Inc talk about OpenStack integration with Compute. Now let's shift gears and talk about network services integration with OpenStack and discuss what Yahoo Japan and Brocade jointly worked on here today. I'm Sandeep Sinkhole, Product Marketing Brocade, and I'm very excited and pleased to welcome Matsuyo-san from Yahoo Japan. Matsuyo-san is the Technical Director of Yahoo Japan and has had almost 13 years of experience working with Yahoo Japan in the data center technologies. We also have with us today Ishiya-san from Brocade's Japan team, and Ishiya-san, other than dropping your iPhone pad, he's also our Business Development Director with NFV, OpenStack, and SDN. And last but definitely not the least is my dear friend, Divya Stove from Brocade Product Management. Divya has had a lot of rich experience with OpenStack and has been involved with OpenStack ever since its inception. Most of you know that Brocade is a market leader when it comes to SAN fiber channel, but a few of you might not know about market leadership in IP networking. We were the first to bring to market Ethernet fabrics, and with our VCS plugin, we provided the first fabric level orchestration. With network functional virtualization, our Vata virtual V router, when it comes to market penetration, is number one considering the number of downloads it has had. Here in Hong Kong, we are showcasing the OpenStack integration with VCS Ethernet fabrics, SAN fiber channel, load balancer, and virtual V router. So Ishiya-san, I was talking to Masio-san earlier in the week and telling him about what I read in Forbes magazine that Yahoo Japan was rated as one of the top 100 innovative companies of the world. Ishiya-san, do you want to touch some more highlights? Okay. So Yahoo Japan is the greatest company, officially and personally in any way. As you see, 80% of Internet users access Yahoo Japan, right? And over 95 million applications downloaded from the Yahoo Japan portal site. Look at that, those great things, right? Why don't we start Masio-san and Ishiya-san today by talking about how many servers you have virtualized and the percentage of these servers with OpenStack right now? Yahoo Japan started server virtualization in 2009 and performance improvement of Hyper Bizer and servers improved VM density for server and our total number of virtual machines will be more than 50,000 in 2014. 50,000? That's amazing, isn't it? I practiced this phrase 100 times. I'm good at it, isn't it? So anyway, for a good reason, which we will talk later in this session, Yahoo Japan adapted OpenStack technology rapidly as from this August. And Yahoo Japan basically abandoned the in-house system management development. We estimate over 80% of VMs will be managed by OpenStack next year. So you guys remember, 50,000 VMs and 80% of that OpenStack, so very impressive. So clearly you're on this fast path for virtualization, but what are some of the challenges you're seeing with the physical resources in your data center? Yes, the number of physical servers have been also included year by year, and this course is efficient use of lack in data center. So as you can see here, Yahoo Japan offers many kinds of services on the portal site, and those services running on the physical servers in data center today. As shown on the left-hand side, the lack fully loaded by servers at the initial deployment, the data center resources are fully available, and lack-space utilization is a high-span service study. However, in a few years, more than half of the servers are removed, and those lack-space utilization becomes less than half. To make matters worse, for some reason, Matsuya-san cannot talk all the details today, but unless the network function is virtualized, those unused lack-space becomes totally a dead-space. As long as there is one single server running in a lack, we can reconfigure the physical lack of switch, top of lack switch, lack of switch. So I'm certain that there are other data center operators that are having this problem. So what are you doing about it? We strongly believe that data center lifecycle management is essential. In order for us to have better control of data center lifecycle management, we need to be abstract data center functionality themselves by making full use of virtualization technology. So if you could abstract all the functionalities in the data center, you are free from all the constraints relating to the physical equipment data center architecture. What this means is that now you can always take the advantage of hardware that gives you less power needed, improved performance, more about the density, and of course much more cost-effectiveness. So absolutely you need to move to these virtual data centers, but how are you enabling the migration and what steps are you taking for that? Yes. In order to realize data center tenant migration, all hardware needed to be abstracted. Right. So in the open-stack world, what abstract hardware resource can be defined as a tenant? What we want to achieve with this is the capability to migrate a tenant from the auto data center to a new data center which is built upon the latest hardware and using open-stack. Sorry. I forget the very important things. We estimate that we could reduce more than 60% of hardware costs today if we could migrate all the tenants which run on hardware that's three years old to the latest hardware. So folks, 60% reduction in hardware costs are very significant number. So the networking service that you have abstracted, can you talk a little more about that? Yes. We started looking into open-stack from multiple angles. In today's session, let's explain what we have done with blockade in the rear as part of my open-stack initiative. In cooperation with blockade, we enable controlling blockade load balancer ADX via open-stack. And controlling load balancer which is essential to web services by open-stack is a very important part of new data center lifecycle management program. Moving to you, Didier, when you heard this business problem from Yahoo Japan, what was the blockade solution that you worked on and can you highlight some of the work done there, please? Certainly, Sandeep. I think that in a nutshell, the very basic, very simple yet very powerful idea that Yahoo Japan has been pursuing all along was to fully decouple, fully unleash if you will, all those application workloads from the physical infrastructure. And as we all know it, open-stack is this abstraction layer in the middle that can really help achieving that goal. So given the fact that Yahoo Japan had already completely taken care of the compute side of the house, it was for them a very natural move to actually take the next step in this progression and to take a hard look at networking starting with load balancing, load balancing being this link between the compute and the networking domains. So when we started to engage with Yahoo Japan, when we started to build our relationship with this great organization, it was about April-March time frame, right, earlier this year. And at that time, what we had at our disposal was the grizzly release. So we already started to work hard and create this brocade ADX driver, this brocade ADX plug-in, ADX being the brocade platform that delivers load layer four to layer seven services, right, and create this interrupt between the brocade platform and this open-stack neutron framework, right. However, Yahoo Japan has very precise ideas as to what they really want to achieve. And they came back to us and told us, look, what we really want to do is to implement a solution that fully supports direct server return. And direct server return is really a great technology mechanism, if you will, that allows the real server to send the response back to the client directly without forcing the traffic to traverse back the load balancer, right. And if you think about it, you know, this is, this solution is actually much more performant, right. Looks like the gain in performance is about eight times the performance that you can get with a non-DSR implementation, okay. So Yahoo Japan, which we love, has very prescriptive requirements, as I said earlier, right. And not only they wanted DSR, but they wanted specifically to have layer three DSR as opposed to layer two DSR, because they really wanted to take the benefits of being able to scale out their network across the entire layer three domain, right, not being bound to layer two segment. All right, so we really wanted to help Yahoo Japan with their open-stack deployment. So what we did, we really started to create a solution for them, right. And again, when we started this project, it was back in April time frame. Grizzly was out there, but as you probably all remember, there is no DSR support in Grizzly, nor in Havana, right. As a matter of fact, DSR, Director of Return, is actually a topic that's going to be discussed this week at the Ice House Design Summit sessions, right. So what is the solution that we came up with? Well, you know, very naturally what we decided to do is to create that functionality within the plugin itself, within the Blockade ADX driver, right, by leveraging this extension mechanism, all right. So we did that, right. We actually delivered to Yahoo Japan this Blockade ADX plugin. The Blockade ADX plugin is fully, is obviously multi-tenant capable, but also supports this role-based access control, which allows admin and users to actually create the load balancer that they need, and that, that solution has been actually implemented, deployed as part of the Yahoo Japan private cloud project, right. So it's been great relationship between the two companies. We Blockade, we actually learned a lot, you know, we learned about how a large scale content service provider, or content provider, is basically managing a cloud computing infrastructure. And I think that, you know, throughout this collaboration, we've been able also to help Yahoo Japan in their journey to full automation. And, you know, we are still working with them in terms of helping them out for automating even more. So as, as you folks might have seen here, that we, we're talking about collaboration, we're talking about innovation. And yesterday, we were chatting about this, that six months ago was the first time when we met Matsuyo-san because of the Portland Summit, the OpenStack Portland Summit. And within six months, we have achieved so much. Essentially, the final thought I would like you to leave with today is that the vendor-customer relationship is evolving, and it's evolving from a transactional relationship to more of a partnership relationship. And OpenStack is that collaboration platform that's driving that, that change in a relationship and, which is helping us to, to innovate and, and push, and drive innovation forward. Absolutely, absolutely. Brocade benefited because we got an opportunity to work on this high scale cloud use case from Yahoo Japan's perspective. They were able to work on a joint solution with the time to market constraints that, that they had. And so it was, it was a match with, with benefits from, from both sides. So that's, that's essentially what we wanted to highlight at this, this session to meet the SMEs from both Yahoo Japan and Brocade. You can stop by Brocade's booth. And with that, I would like to, on behalf of both CSUN, Didier, I would like to thank Matsuyo-san for taking the time for sharing this user story with us and, and thank you for being here. Anyway, we can show the live demo at the booth, right? Absolutely, absolutely. 300 VMs? So we have, so we have the Yahoo Japan team at, at the booth and we, we both have, we have a recorded demo and also a live demo that can be shown at a booth. So stop by, watch the booth and, and we'll be happy to answer the questions that you have. Thank you so much. Thank you. Thank you. Thank you very much. Thank you very much. Thank you. Thank you.