 Yn ymweld i gwasanaeth yma ydych chi? Mae gweithio'r ysgol yma i gweithio'r problemau a ddim yn gweithio'r ymddangos rydych chi'n credu fe yna. Fe yw Peter Agnew, dyma'r cyfnod cyllideb yn gyfnod cyllideb dyma'r cyllideb gyllideb yn cyllideb Cymru, yw cyllideb yma'r cyllideb Gwyddoedd Cymru, yw'r cyllideb yn cyllideb gyllideb gyllideb gyllideb. Mae'n cyfnod, mae'n dweud hynny fydden nhw o'r cyfnod o'r cyfnod i gofynu'r cryn o'r cyfnod i'r cyfnod, i'w dweud hynny'n dweud ychydig o'r cyfnod i'r cyfnod i'r cyfnod i'r cyffredinol, felly mae'n dweud ar gyfer y prototypet a'r ffant yw'r ffordd i'w gwybod i gynnwys ar y cyfnod, os ymdweud o'r cyfnod i'w dweud hynny a'r cyffredinol i'n dweud y dyweddig a'r cyffredinol i'n dweud hynny, Ac os yw'r eich drwy'n dyfodol gyda'r cyd-ddoch nhw? Dyma'n fydd ychydig arweinydd, a'i ganddech chi'n ffordd teithio'n gwybod wrth dod o'r rhai囝cio o'r cystafliau? Fyddwn ni wedi'u gwiaith o'r ffordd o'r dweud, ddaeth fyddwn ni'n meddwl, dyma'r gweithio y cwestiwn. Rwy'n credu y gallai'n cael mynd i mewn o gylluncol a'r yrhyw ychydig a'u cyfreidd raenio? Dyna ddyn nhw ddadiwch chi eich anseid dylai, I will provide the answer. The link between them is this man called William Stanley Jebbins. He was an economist in the 1850s in the UK. And what he observed was that as the factories and the industrial revolution grow, they couldn't get the coal out of the ground fast enough to serve them. So they used technology to make the consumption of the coal more efficient. What he actually observed was the more they used the technology to make the consumption of the coal efficient, the more coal and totally consumed. So now, how does that apply to technology services? If you want to have a read that as the actual definition of it, how it applies to many things you can check in Wikipedia. So, our talk here is where the inefficiency is, and it's about OpenStack and its role in the sort of one-aware multi-cloud. If we look at the environment of cloud enterprises and the wide area network between them, the IT environment for a long time has been working on abstractions between different things, very well defined. What it means is that cloud environments are very, very efficient, very, very elastic and very, very quick. However, the wide area networks are pretty much misaligned from this. They've got a lack of abstractions and they have long lead times. Now, software defined networking is starting to move some direction in that, but maybe not quite there yet for the wide area network. Come from the enterprise, a direct quote there from an enterprise that they can move, they're virtual data centres from one data centre to another, or from one cloud environment to another, in a matter of minutes or hours depending on the amount of data, but it takes them three weeks to get a network changed. And that is where the problem lies, is in the networks. So, if we can actually increase the efficiency of how they actually use those networks, then we'll actually increase the consumption of cloud services within that environment. And it's really down to the issue within the wide area network and how it is slow and it needs to move faster. Why do we care? Why do cold care? We actually sit in the middle of a whole bunch of customers, data centres and buildings. We provide fibre optic connections directly to over 20,000 major buildings in Europe. These are large enterprise buildings in city centres. We also provide those to the 450 or more now data centres where the cloud services where people meet cloud services. So, if we can sit in the middle of that and make it more efficient for our customers to actually use these services, then that is good for us and the customer. So, the bottom line is, if we can make the inter-cloud one more efficient, enterprise IT will consume more network and more cloud services and the more efficient we make it, the more they will consume. And the easier it will be for customers to actually run their IT services in a cloud environment. And hence actually start to use even more IT services according to Gavin's principle. So, to move on from that, I'd like to pass on to my colleague from a sister company, KVH. Good morning. Good morning. My name is Caleb Crane. I'm the director of software development at KVH. That means that I am in charge of the public and private cloud at KVH, developing the software for that, as well as the customer portal that we give to our customers. That said, KVH is not a software company. We are actually a telecommunications company. We provide network services, data centre services and managed services to our customers who want to do business in Asia. We connect 22 financial exchanges. We connect more than 140 data centres and more than 2,000 customers. As you can see from the title of the slide, we consider ourselves to be the pan-Asian information delivery platform. That's corporate speak for we provide network connectivity and data centre services to all of Asia. But we don't just stop at Asia, we actually provide connectivity back into the US and then through our sister company Colt into Europe. So, anyone who is from Europe who wants to do business in Asia, then they have a good time going through us and the same for the US and vice versa. So, I mentioned a moment ago that I'm responsible for the public and private clouds at KVH. We first built public clouds at KVH in 2010 using our own custom orchestration management software. We provided bare metal provisioning, virtual servers, layer 2 networking, firewall services, load balancer services, any kind of infrastructures of service that a customer could want. We like that platform. It took us a while to build it, but over time it's become a bit of a bear to maintain. And we've also been keeping our eye on OpenStack and we feel that OpenStack is catching up or exceeding the capabilities of our custom system in many places. So, I'm also responsible for transitioning us and our customers to using an OpenStack-based environment. So, for us, what does OpenStack mean? Beyond just doing cloud provisioning, we think that it can also do cloud federation. And it can be a core component in enabling an inter-cloud when automation. As you can see from the slide before, KVH is not just a cloud provider, we are a network provider. So, for us, the cloud automation is important. However, the network automation is as much if not more important. So, we think that OpenStack can be the base of a platform that will enable us to do all of that. For that platform, we will provide a single customer portal through which customers can access their KVH private clouds, but then also occasionally, or maybe frequently, burst out to use services like Amazon, SoftLayer, whatever public cloud provider is popular that week. However, when they burst out, they also need to have network connectivity. They need to have connectivity between their private clouds at KVH and those public clouds. At KVH, we provide many different types of connectivity, and we would like to actually federate not just the cloud, so that when customers use the private cloud at KVH, they will use the public cloud, they'll have one interface, they won't have to go to Amazon and then to SoftLayer and back to KVH. They just go to KVH once and use us and everybody else. We would also like to federate the networks that we use to provide that connectivity. For us, an orchestration system that takes into account not just the cloud, but also the WAN, is extremely important. Like I said, we think that OpenStack can provide the base framework for that at a very high level. Again, that looks a lot like this, a single interface that a customer uses to control their private clouds, their public clouds, as well as orchestrate their network connectivity between them using either a direct connect from AWS or a direct link from SoftLayer or whatever they're using on that day. They should not only be able to provision links, but control the bandwidth of those links. So I want to have 100 gig today, 10 gig tomorrow. I want to reduce the bandwidth utilization using the customer portal. So all that's well and good, but it's pretty complex to interact with all these different systems. That's where we're looking for a single, as we mentioned before, federation layer or a single layer that will do all the automation for us. We believe that that can be provided by our friends at CRS, and they've made a lot of progress on that, and so I'll hand over the presentation to SIG, and he will walk you through the CRS platform and what they're going to provide. Thanks, Caleb and Peter. My name's SIG Lwft. I'm the CTO and co-founder of CRS. I'd like to spend a few minutes to talk about our platform and give you some ideas as how we're looking to solve these problems. CRS is developing a platform that we call Cloudscape. Cloudscape is a software platform that is targeted towards the service provider market. Its main objectives are to provide a single interface into both the service provider's private public cloud assets, manage the network of the service provider, and as well leverage the public clouds that everybody is ultimately looking at in some way, shape or form. We believe the service provider has a very unique position in the cloud evolution. One of the main reasons why we think service providers will play an important role is the fact that it is the enterprise who's going to cloud that already has a relationship with the service provider. As both Peter and Caleb pointed out, they already have tens of thousands of customers and that first mile connection is mandatory if you're going to do anything in cloud. You can't use cloud if you can't talk to it. A number of applications that are reasons motivating enterprise to go to cloud are the basic cloud premises. I don't want to buy a new server every 18 months. I'm tired of paying for all of this capex. I'd rather go to an OPEX model that you find in the cloud. If you take an application in the enterprise and move it all the way into the public cloud, there's risks that can come with that. If you just think of email, probably the best place from the enterprise point of view might be here as opposed to in the deeper waters of the internet. At the end of the day, what the CRS platform is looking to do, however, is to leverage the network, leverage the data centers, and leverage the public clouds for the benefit of the service provider. From a high level, we like to look at this as a sort of a simple problem. There's cloud, there's WAN, there's applications. It's really the applications that drive everything. And as people go from single data centers in the cloud to multiple data centers in the cloud, you end up coming to a scenario where you need connectivity. Not all applications need dedicated connectivity, but there's a lot of mission-critical applications that require it. And for this picture to work, the applications need to talk to cloud orchestration and they need to talk to WAN orchestration. Generally speaking, it should be a simple problem. The reality of the world is not as simple as you'd like. There's a variety of technologies out there. OpenStack certainly is making some big headway on the WAN side. SDN is certainly improving the situation, but it's not a completely stitched up story today. Ciaris tries to play on both sides of this line. Our mission is to make the problem simple, allow the applications to do what they want to do, make sure that the cloud orchestration and the WAN orchestration are working together smoothly. And we do this with an OpenStack focus. We believe heavily in OpenStack. We think OpenStack is going to be the dominant cloud orchestration. And ultimately what we want to do is we want to let applications accomplish the goal that they have at hand and do it through an OpenStack interface. As well, we've talked about focus on service providers. Part of the story is to make sure these applications that are being deployed in the service provider can get the connectivity they need. But where that connectivity is going out to the public clouds and it's a very reasonable use case, we also need to be able to make sure that that connection goes all the way through. So, not only are we providing the ability to deploy cloud data centers into an Amazon footprint, we also are driving the connectivity, the direct connect connectivity to go in concert. So, if you want to burst out that cloud for a couple weeks, you don't only get the cloud, you get the network at the same time. And just to fill out the picture, as mentioned by Caleb, they're looking at soft layers and other potential public cloud. We're doing some work on Azure right now and ultimately the goal is to fill out a broader spectrum here. If we take a look at our architecture, and I won't spend too much time, we don't have that much to talk about, but happy to answer any questions you have afterwards. Our platform, as you can see, is built on OpenStack. To the northbound of us, what you see is OpenStack. Interface is coming in to Cinder, Neutron, et cetera. We've added a few things to the picture, however, to address the problem of WAN. Core to the center of this is the introduction of what we call a WAN project. And this allows us to build two additions to Neutron when we call, which we'd like to propose to Neutron, when we call traffic engineering as a service. And this essentially allows data centers to engineer the traffic from application to application between different data centers. The second piece is we call WAN as a service. And really the objective of WAN as a service is to allow connectivity to be established, dedicated connectivity to be established between different data centers. The connectivity itself, what might appear to be a simple request I need to go from here to here, actually goes through a variety of network types. And as we already mentioned, Direct Connect for Amazon, for example, or ultimately we're looking heavily at the SDN world to evolve as it's planning to. Other things to comment about our architecture is the ability to have data centers projected into different types of clouds. And when you think of OpenStack where you develop to an API, so you have a Nova API, you do a driver, driver maps into underlying hardware, we do something unique where instead of mapping it into hardware, we actually map it into what we call an undercloud. And the undercloud essentially can be any type of orchestration platform and whether it's private or public. Obviously we prefer OpenStack, but we can support VMware, we can support Amazon and are developing towards other public clouds. So with that, I'm just going to sort of summarize. What we see is the network is becoming an important piece of cloud. As people are going from this traditional test dev and looking to more commercial grade deployments into the cloud, issues such as networking are coming up. And one of the biggest sort of disconnect between cloud and WAN is the fact that WAN oftentimes takes weeks and months to be provisioned and it really doesn't fit the cloud model where you want to plug it in, or not plug it in, just plug in your credit card really, turn it on and run it for however long you want. So our whole goal is to align the cloud and the WAN so that you can deploy the two in the same way. And we believe that companies like Colton KVH can greatly benefit from this type of capability. So with that, we'll open it to any questions you might have. Okay, well if that's the case, thanks very much for coming.