 So, I am talking about dynamic infrastructure and cloud computing. So, I am sort of approaching cloud computing from a very different direction and let us see if we manage to reach cloud computing by the time my time runs out. So, this is sort of a high level thing to start with. So, we know about the world is getting smaller and the world is flat and all those things it is also getting smarter in many ways. It helps that we are getting smarter I think, but it is also other things in the world that are helping make the world is a whole smarter. So, what do we mean here? There is a lot of technology going into day to day stuff right. So, it is not just about us developing technology in the labs or in universities and so on. It is also about applying the technology in mundane things like managing a traffic, managing water supply, managing electricity supply and so on. Things in our day to day environment are getting smarter because we are instrumenting them because we are putting in the ability to measure what is going on, to gather a lot of data and make sense of that data and use that to improve the way these utilities are delivered for example or improve the way things operate in general in the world right. So, there are many examples of that out here, oil field technologies and you know supply chains and things like that. So, like I was saying it is basically all about instrumenting the world and the amount of instrumentation that is going in is phenomenal right in various forms. So, as it says out here the world will be 10 times more instrumented next year as it was in 2006 and that number is actually going to something around 1 trillion instrument you know sensors essentially that are gathering data for us and there are sensors of various types I mean there are the more obvious ones that you would have heard of like RFIDs and so on. But even things like you know your setup box which gives you satellite you know dish TV and things like that. That has the ability now to record stuff to understand what you are doing with channels you are watching for how long and so on and that data can be collected can be used can be analyzed in various ways. So, there are many examples of that as you see out here medical imaging you know all kinds of data being captured as part of the tests that we do in a hospital right there is of course, privacy implications out there as well, but it is useful to be able to access that data when we really need it. But the implication of course, is that there is a huge amount of data being generated and there is a huge amount of computing that has to be applied to that data if it is to be useful right also as it says out here something like 70 percent of the digital universe currently is created by individuals. However, the job of managing that data falls upon the enterprises that help you to store the data and to analyze the data and so on right. So, enterprises are responsible for 85 percent of the implications of all that data of managing the data compliance and security and all those issues right. The net result of this right now is that you know these relatively small number of enterprises that are trying to manage humongous amounts of data on behalf of the hundreds of millions of us or may be billions of us who are on the net in various forms is that the IT infrastructure that is being used to manage all this is at a breaking point. And there are various you know data points out here you know towards that effect. One of the things to note here is that you know as far as the spending is concerned on IT what companies have to spend 70 cents out of the dollar 70 percent of the spending is basically just for trading water if you are staying where you are making sure things keep running right just maintaining what is out there and that is too high because that means we are not we do not have enough budget left over essentially to expand what we can do with all the IT that we have right with and to keep up with the growth in the data the growth in the applications the number crunching that is needed right. And there are various other things like that we have already talked about idle time in all these data centers right and all the computing infrastructure that we have because of inefficiencies in how it is used typically you would find 85 percent idle time on many machines even the best managed data centers can only get it down to like 80 percent 70 percent those kind of ranges. So, obviously you know it is partly the IT infrastructure is at a breaking point because we are not able to manage it well enough because we need better ways of running our data centers and so on ok. So, that is why it is sort of time to start thinking a little differently about the infrastructure and the point being made here is the infrastructure itself needs to become more dynamic in various ways I mean it is although we focus mostly on IT infrastructure by the way this also applies to other kinds of infrastructure that we use. So, it could also apply for example, the telecom infrastructure right communication is so integral to everything that we do in IT as well not just in our daily lives. So, even those kinds of infrastructure or mobility infrastructure whether it is you know trucks carrying goods all over the place or whether it is people being shipped all over the place in trains and buses efficiency in that infrastructure is also critical right. So, there are all kinds of infrastructure that needs to be managed in a more dynamic fashion and IT can help to do that that is you know basic message here. So, the three things we need to consider out here we already you know I am sure you are familiar with virtualization. So, in order to make the infrastructure more dynamic what do we need we need virtualization why because we can then share the infrastructure more easily. Virtualization lets us partition our resources into finer granularity chunks which can be shared thereby improving the utilization of the infrastructure as a whole. In order to do that we need standardization right because I have a bunch of IT workloads that I need to run if I am going to be able to run that on virtualized infrastructure and then move things around then you need standardization of the platform underneath. So, that your applications can execute on different platforms and dynamically change based on the demand and you need automation because without automation managing this sort of a complex infrastructure is just is next to impossible. So, it is hard enough for enterprises to keep up with managing their IT infrastructure as it is currently. Once you make it virtualized and once you start loading servers up with more and more you know heterogeneous applications which have different failure modes and so on. You have an even bigger management problem to deal with and then humans you know quite simply run out of the cognitive capacity to deal with those kinds of complexities. So, you need automation in various ways to help the human administrators to help the folks who are responsible for keeping things running because otherwise without automation without tools that help you to run your data center. There is essentially no hope of scaling beyond what we are currently today and when you combine all these together the net effect is that your cost can go down and your flexibility goes up right and we have seen this previously in history and there are other examples of this it is not just about IT right. So, we have seen examples where you know in the old days there used to be telephone operators plugging wires into the boards right that obviously got industrialized it got you know automated. So, that we have automated switching now and similarly you know manufacturers. So, even the model T the Ford model T the classical example of great innovation by putting in an assembly line and speeding up the production of automobiles even that still required humans to be putting the parts together has gone away as well in most modern auto companies you will find that this all robotics has putting the car together right. So, there is industrialization happening in various different domains banks I mean we are all familiar with this right long queues in front of tellers have been replaced by short queues in front of telling ATMs some progress at least. So, it is similar with IT there are different kinds of workloads and we need to sort of analyze the workloads and decide how we can you know dynamically adapt to serve different types of workloads. So, the characteristics of the workloads are what will sort of determine the rate and the standardization of the underlying IT services that support them right. So, there are different types of workloads that we can consider there is analytics kind of workload I was talking about the huge amounts of data the need to analyze derive patterns out of it get business intelligence out of the data things like that very different kind of workload is your collaboration workload you know large teams working together on projects geographically separated and all that they need tool support in order to collaborate in order to communicate with each other share document share data things like that different workload, but nevertheless an important kind of workload to support. And then similarly the others like you know development and test kind of environments or you can have you know at a simpler sort of level of sharing you can have compute sharing and storage sharing as well. So, that you put your you know data on the cloud rather than having to have it all in house right. So, essentially the idea is to try and provide as much of self service as possible to these end users who are using these IT workloads on the IT infrastructure right. So, you start with a self service portal of some sort. So, that the end user can directly come to a portal and access services in as automated a fashion as possible. We are trying to minimize the need for IT staff to be to have to support these services and to be in the loop so to speak right. So, essentially you come to a self service portal you raise the service request in response to which the service gets provisioned the necessary resources get allocated automatically in the background in response to a request by the user at a portal. And all this can be automated at this stage right. Of course, as of today enterprises or service providers are not yet comfortable enough with the idea of complete automation which means that they would still want some kind of an approval phase in between out here. Before the infrastructure actually gets provisioned and given to the user to use they want to have somebody you know controlling the gate and saying okay to go ahead or not okay to go ahead based on various business criteria. So, they could still be a human step in this workflow but it is still an automated workflow. So, that human who is monitoring what is going on in the data center and deciding whether to offer this service or not to that particular end user only has to click a button on a console or maybe just you know within an email you might have to click a link or something. So, these workflows can be automated so that it is quick and easy rather than having to you know rather than having the human be a sort of bottleneck in the whole process. But essentially you are getting a whole lot of automation that lets you create a dynamic infrastructure to support all of this which brings us to cloud computing in a way right. Because now in order to satisfy this we need the ability to dynamically provision resources and not just the underlying resources in terms of compute and storage but also the application of the services on top in response to user demands in response to what the user is asking for from a self service portal of some sort. So, I do not have to go into the details of you know what public clouds are and private clouds are presumably that is pretty obvious for now right. I mean so public clouds so I mean there are there are it is a spectrum of possibilities out here right. So, depending on the scale of your enterprise depending on the scale of the services you are trying to offer you might be ok with a private model a private cloud model which is purely held inside your data centers or you might also prefer to go public in a sense and borrow resources from a public provider or lease resources from public provider as many need to right. And there are various considerations that come into the picture when deciding which models to use in the cloud. In many cases security and privacy considerations could dictate the choice of a private cloud rather than a public cloud as of now. And these things help basically to bring down the unit cost by enabling economies of scale. So, if you are if you are doing it the traditional way hosting all your infrastructure yourself in your data center then the unit cost taper down only gradually right as the scale grows. You can build an internal cloud and bring it down much more significantly essentially because you are sharing you are increasing the sharing of resources using techniques like virtualization server consolidation and so on. And then for the service provider clouds the public clouds the scale can you know this is this is started tapering much much further behind because basically you are operating at much larger scales and you derive those economies of scale. Well it is a very gradual curve like I said it started tapering way back somewhere right. So, typically we expect the service provider to operate at much larger scales than a typical enterprise. So, there are various kinds of cloud based services that can be offered and I am sure you have seen some variants of this before. So, they can be infrastructure offered as a service which is basically compute or storage kind of infrastructure which can be rented on demand. There is platform as a service which we already heard about right the Azure example or Google App Engine and things like that. And then there is software as a service right. Now cloud computing is starting to encompass all of these things and sort of being used as a catch all term for all of these. But really our focus so far primarily has been on infrastructure as a service especially within IBM at least although we also have offerings in each of these layers right. So, just in terms of terminology basically trying to define that the cloud is really here in the sense of you know offering the infrastructure itself you know in a form that the customers can use without caring how it is allocated how it is managed and so on right. But typically the users of the cloud are the ones which are offering services on the cloud or offering software as a service. So, this software as a service I think we should kind of leave it alone in a sense and it already there is already a good term for it is already a good space for it. And there is a different set of challenges in creating software as a service and an offering software as a service right. And then there is platform as a service in between which again is a sort of niche in between basically because you have somebody who needs to create software as a service but does not want to be bothered with managing all the infrastructure underneath right. So, it is an in between layer and it is an open question as to how long it will last in a sense. I suspect that you know in overtime parts of the platform as a service will gradually merge either into the infrastructure layer or to the software layer right. It is a development platform essentially that is what it is. Underneath of course, in order to build infrastructure as a service you need platform products. So, you need the hardware and the software necessary to create the infrastructure as a service layer right. And there are some examples of that out here. And all the way on top there is again something that obviously IBM is always keen to project which is services right. Doing all of this is not straightforward. And therefore, there is also expertise that can be offered as a service right. How to build cloud applications, how to build SaaS applications and things like that. So, now this is an attempt to sort of you know show that there can be a slightly different approach to looking at cloud computing in this case. Which is that you align the choices that you make the kinds of cloud deployments that you consider based on the workload that you are trying to host in the cloud right. So, do not worry about the specific examples out here. But essentially the idea is that you know depending on the kind of workload you are putting into the cloud you may have a range of choices of different kinds right. So, at the bottom is basically the systems, the platform components which let you create clouds easily. Then between you have you know private clouds and then you have public clouds or standardized services offered on public clouds right. So, what is appropriate, what is the appropriate choice might depend on what workload you are trying to host on the cloud. So, in the case of you know things like compute and storage infrastructure it is perfectly fine to go with the public choices. And this is what Amazon is doing right EC2 and S3 basically. So, the compute cloud, the storage cloud if that is essentially what you are trying to request from the cloud. Then it is fine to go with the public cloud and create your own layers on top right. But there are cases in which you might want to you know move down to the lower layers in a sense either to keep things private within your data centers or because you know there are prepackaged kind of capabilities that you would not get in a public cloud. The public cloud by default has to sort of serve the least common denominator of the customers right. I mean the request that everybody is likely to ask for just because of economics right. They have to be able to scale. And there are times when you cannot offer you know complex focused services in a public fashion because there would not be enough big customers to justify that kind of investment. So, for example, you might have something called an analytics cloud which offers the compute capabilities, but also offers the layers on top that do smart analytics of various kinds business intelligence of various kinds right. So, essentially you could and secondly when you are using something like that you do not want to expose your data to the public cloud. This is you know you are trying to derive business intelligence out of a collection of data. You do not necessarily want to put that data on to the public cloud and take the privacy and security risks of that. So, you might want to have a hosted cloud a private cloud within your own infrastructure, but be able to use that flexibly in the sense of again you know these workloads will vary over time. You will only occasionally have to run your business intelligence workloads. So, you are underlying infrastructure can be shared and multiplexed across multiple workloads. Whereas, as and when you need the analytics workload to run you have the cloud layer available there to run on the infrastructure which can be obtained from your local private clouds right. And then there are the kinds of workloads. So, you know for example, for collaboration again it makes sense you know everybody out here I am sure uses web based email right. So, it is software as a service it is public and we seem to be comfortable with doing that even with our private email and so on right. So, depends on the acceptance levels out there, but basically yes we can have hosted you know public clouds hosting collaboration tools and we have examples of that out here or you could do it in house. So, you can have a pre package sort of appliance which lets you create a collaboration infrastructure in house again using an underlying layer of flexible resources compute and storage and so on. So, there are many such examples out here of you know of workloads being mapped to you know based on their characteristics based on the sort of business requirements for those workloads mapping them to the appropriate layers in the cloud in you know matching them to the appropriate cells out here in this grid essentially. So, to summarize you know we sort of approach this from the point of view of a smarter planet that we are trying to build out here right. So, there are a lot of opportunities that are arising because we are instrumenting so much because we are instrumenting all our system so much and the net result of that has been that you know there are newer kinds of IT and business services that are needed they are emerging because we are generating all kinds of interesting new intelligence out of the world around us right. And so, it needs the underlying infrastructure it needs service management systems it needs the ability to deliver infrastructure to manage all this data and computing requirements in an automated intelligent manner right. So, there is a whole lot of work needed in service management and when we say service it could be IT services it could be other kinds of services as well, but those services have to be managed and delivered in an efficient manners if we have to have any hope of scaling it to the kinds of you know scale that we are now considering like a trillion devices connected to the net and so on. And then these you know the cloud model which is basically more of an IT consumption and delivery model right at the layers of the infrastructure and you know infrastructure software and hardware. Those are already very compelling for some workloads and we are in the process of sort of gradually testing the waters here with different types of workloads testing out what kind of impact it has to be hosting your workload on a cloud like infrastructure. And so, that will evolve over time and I already mentioned you know there are various layers at which you can have various offerings to let customers dip their toes in the water essentially and try things out gradually. So, that you do not suddenly sort of start off with a public cloud and try to host your mission critical applications in public cloud nobody is going to take that risk, but you can have gradual sort of graduated steps leading towards that starting with flexible sort of appliance models within your data center then going to a private cloud kind of model within your data center. And then gradually and in a controlled fashion outsourcing your infrastructure needs to the public cloud that is all I had Thank you. Questions? Yeah, nowadays I think we also started hearing about sky computing I mean can you say something about it? I haven't heard of what sky computing. So, how is that? Across clouds in fact. Okay. So, there is there has been some effort to try and sort of standardize the interfaces to the cloud. So, that you know so that you are not stuck to a particular provider of the cloud. If you build on a particular platform, if you build on Azure, if you build on Google App Engine or whatever then essentially your applications will only work as of now on that cloud. Some customers are not comfortable with that right because they want the flexibility tomorrow if they fall out with Microsoft they want to be able to move away from Azure. As of now there are no standards there is just some initial efforts at standardization. As and when that comes about you know if that comes about I should still put that if out there. Then certainly yes there will be the ability to you know migrate workloads across clouds across cloud providers. But frankly that's too far in the future right now and I'm not too sure that it's a complex technical problem to solve. It's just a matter of agreeing on standards as far as the API layer is concerned as far as the platform layer is concerned. Sir, what is the scope of cloud computing in manufacturing industry and sector and how cloud computing and dynamic infrastructure will be helpful to make a smarter the industry. Okay, so in the manufacturing industry just to give an example I mean these automobile companies right they have to do a lot of number crunching as well. They do a lot of simulations and so on when they're doing especially in the design phases right and again so it's more of compute workloads that they have to host. Basically in the workshop like when CNC machines and other machines are operating then how it will reduce the cost of the industry. I'm not sure that cloud computing directly address you know the CNC machines and the assembly process itself. But there are other ways in which they end up needing IT infrastructure which can certainly be virtualized and shared and so on. Not sure if cloud computing can address the actual sort of manufacturing process. Thank you sir. A high value business transaction say banking your large database base you want a secure commit happening that kind of a workload is more of otherwise the compute workload where you want a huge amount of a design kind of thing. So you got large databases doing a lot of transactions you want an update to be committed. So there are business reasons why you would typically not want your part even if you set up your own private cloud. Is this technology really amicable for that. I have a four or five applications running. Each of them are a critical transaction independent application. They happen to be a common what you call a platform infrastructure. Right. Is it feasible for me to have a ubiquitous a farm of a machine. A layer of a database service. Absolutely. Is it feasible. Certainly within your data center you can have a more flexible infrastructure that way. You can have a farm of database service. So is it feasible for us to one and define a business rule kind of a rule engine that if the utilization goes in a one server beyond a point a kick of a second server or a do the V motion of the entire image from one to other is easy to do that. Right. The same sort of graphs that Varsha showed right would apply also to transactional kind of workloads. So depending on the kind of transactional workloads you're seeing you could certainly scale up the infrastructure that you have at the back end. The database servers that you have the back end. You just mentioned there are really no standards today. Let's look at simple things. Virtualization basically is core to build any cloud. Right. They are competing virtualization technologies. Even that is not a standard. They're not standard. So you will not be able to move an image of a working image and a Linux guest always working the same VM where it was a reddit virtualized machine. There are no way to negotiate that I would like to transfer image from a once people offering me So there is there is some progress on that front at least in terms of virtual images and so on. There is something called the open virtual format OVF which lets you do some mapping of images across some platforms depends on who confirms to what standard to what extent and so on. But that there is progress happening there at least. Especially with reference to the private clouds that you mentioned. Would you consider private clouds as a means of providing IT service management as one means of providing the IT service management company. Yeah. So you look at cloud computing as a means for the IT service management. So the two aspects of this one is the cloud computing itself. I mean the cloud itself can be used to host your IT service management applications. So there are multiple automation kind of tools that are needed. Those could be part of the cloud. So the superset actually would be IT service management and clouds would be a means to an end. Sure. Clouds would be one part of delivering IT services. I mean IT service management will encompass clouds as a way of sharing your resources efficiently. But it would also need the other automation tools and so on to keep the cloud itself running in an optimal fashion and things like that. Thank you.