their web sites. what you do not know is one minute after the doctors at man -- multimillion-dollar advertising campaign, the website fell down. you did not want to be the i.t. guy there, right? furthermore, is widely believed that the problem that it was found on the denise site a week before. for those of you not technical in the room, i will tell you, the problem could have been solved technologically, meaning you could generated synthetically a ton of load coming from multiple sites around united states to simulate every case i just told you about. you technologically could have done that. economically, it would have cost you an arm and leg to do this. which is why nobody did it.

today, a whole host of guys are saying, would you like 10,000 computers for three days to generate a ton of synthetic load and hit all these giant websites? you could have that you only have to buy it for three days because they have delivered -- amazon among others has done this. you want to beat a student of this, it to be a student of business models. they have a concept called reserve pricing, meaning if you think you're going to use it for a year, they a little but up front, and they will take the hourly rate down. then they have a fascinating thing called spot pricing, which means that every hour, we can all go bid on whatever excess capacity there is. each you are the high bidder, you get it. there are a lot of interesting applications for this, by the way. this innovation and business model has not stopped. they have let this thing, but it is not over. that, to me, is the holy grail.

this technology is going to enable people to build business models which you have never seen an economically allow things to happen which we have never seen. is that a short answer or long? >> that was just right. thank you. >> my next question leads into that with simon. you have developed some core virtual is asian technology. you have seen it be used in private clubs, in the enterprise setting, in public clouds. how do you see customers reacting to both private and public clouds, the challenges, the opportunities? >> [inaudible] because of the scary things out there, and because they have real humans who are scared of letting go of control. again, it is not so much a technology challenge as a human process challenge. it is really scary to let go of this thing you have been charged with looking after. it could be a regulatory

environment, but it could just be that your job is built around regulate -- making this thing work. so we have a skill set issue, too. in general, they are building protocols for security reasons and a bunch of other things. in general, it is absolutely in existing i.t. people's interests to build a private club because they get to keep their jobs. so they are off doing that. but the broader concern that is articulate it is one of security and control. and the public clout is been generally branded by the enterprise segment as being untrustworthy unless it is backed by a particular promise to have some major brand like microsoft or google or whatever. what has been lacking has been standards related to a customer of visibility to reason about the cloud. but we have some good anecdotal evidence that people will focus on a cloud for a living will do a far better job than you would

ever if you were building your own i.t. system. a good way to think about this would be if you look back on wikileaks and what followed, it turned out they were hosting on amazon, and when amazon pulled that, anonymous launched an attack on a bunch of different web sites. everybody who was running their own data center was down in a heartbeat. so statement one -- if you are running your own thing, you will be taken out much more easily than if you are running as part of a much larger aggregated whole world that provider can defend the property much more rigorously. and goodness me, they care about defending it. but more importantly, through automation, you get to get an assurance that data is less accessible. so people will say to me that if they build a private cloud and surround it with people, they

are much more secure. that is patently false because people tripped over cables and things fail. but people also have usb cheese and walk off with your data, and this is highly known. again, the data should never have been there and the guy should not have been able to download it. that is a human problem. a failure human systems. there is alerting and people not looking at a human log. humans are bound to make mistakes. when we leave it to humans, humans are inclined to believe -- inclined to seek fun. when we look at the security angle, security designers always believe that users are prepared to trade security for functionality or just for fun. you know what -- i tell you what -- if you get a user advancing -- if you give a user a dancing

paper or something like that, they will click on it every time. problem is we are gullible and render our enterprises are vulnerable to attack. human nature of most of the attacks are actual so's your engineering. so, again, the challenge is the your private cloud might be well-fortified. in general, attacks come in to the data center generally on the basis of compromise clients. so in the google attack, for example, someone click on a poisonous explores six url -- a poisonous internet explorer 6 url. the next set beyond getting out of the infrastructure business is to get into the business of trusting a vendor who already owns the facilities and builds servers and is reinstalled with everything to make them

automated. let's call that a host of private club. the model talk about some dances called hybrid club. we already have this setup, and we have a hybrid club were some of it is run by somebody else, and that is there clearly were the enterprise is aiming for next. in general, the enterprise has to adopt the cloud. the moment you want to access an application running outside your parameter, it is not clear what the users identity is. enterprise is exploding both at the front and the back. i mean we all want to sit in starbucks and work, so the enterprise network is no longer there. we're all accessing stuff over the public internet. and then at the back, the enterprises adopting these clouds services, whether it be private computing for the challenge is at the back, the enterprise has a vested interest based on identity. so i, sitting in starbucks, want

to go to -- was just a salesforce that, -- let's just say salesforce.com. whose identity am i doing that under? the whole problem is that security is tighter i'd be. the moment you step out of the enterprise boundary, the identity is up for grabs. so extending the enterprise boundary, extending security, extending the enterprise into the cloud is a major challenge. they have to extend not only identity management, but also all the requirements for security and clients and those sorts of things. so the adoption of various cloud services notwithstanding, it is growing at pace. there are major changes still technologically to get it done. so the enterprise use public clouds as untrustworthy, and reliable, and scary, and yet, every single enterprise employee uses it. on the enterprise client, and

all their own. and of 40 walks in with their own network nowadays, and the process enterprise data on the rise, and they probably -- it probably drag it into something like dropbox. tim mentioned an interesting statistic of how many people said they were using cloud. the only thing that is interesting is cio's who do not know that they are using every form a cloud that is available. this is the charge again. if the employee wants to be productive, a well-intended employee wants to be productive, just to send an e- mail to a colleague, and bounces off, so they drag it in dropbox and send it publicly, and suddenly all your data is in the cloud whether you like it or not, and the cio has no clue. the consumerization challenge is the challenge for adoption of the cloud. broadly, from an automation perspective, it is the case that for the omission build you a much more reliable concept.

i happen to be a passionate fan of the models being built because their entire business depends on doing a good job. so you can bet they will have far more highly incentivize people, but fewer people, and lots more software. so i appreciate that perspective. as a practical reality, we see customers come to us every day. >> one of the very nice things is you can move consumption from this model to a consumption model, which is really important here that is a key change. that you no longer have to buy stuff and stand it up and pay for software licenses. you can simply read them. that is a phenomenal change because you're the consumption- based. >> which has been my final question to all three panelists, which is where of the future of cloud computing going?

will this be an instant gratification engine where, as you mentioned, the power of the internet is at the hands of a particular individual, or is this a way of life? >> i think it is a way of life. we are on the second year of the 20-year wave. i think 20 years is conservative. i think this is what we are moving to really as an information technology initiative. this is what the future holds now for us. i believe that the time of infrastructure management and procurement of infrastructure, that time is surely waning. it is waiting today. if you look at some of the budget models and some of the forecasts of some of the perennial infrastructure companies. they will all tell you that sales of their pc's and servers

have decreased significantly just within this two-year time, so if we are making this type of progress and growth just in the past two or three years, what happens year 5? what happens year 10? i think to your earlier point, many of the common, everyday services and tasks we do every day will become -- will all be in the cloud. so much of it is today already, but i think much more. i think one thing that will continue -- it will actually reach into the home more than it already has. i think that is where the next advent of this will be. it is already pervasive in business. i think it will in some way -- into the home -- reach into the home, to the consumer, touch the consumer within their home. >> i predict that employees will be bringing their own devices.

just as you bring a 10 in to write, you'll bring in your voice, and it will be an expectation that employers will stop writing enterprise devices in the not too distant future, simply because there is no reason for them to do that. then, from the adoption of cloud computing perspective, for sure, it is growing fast. it is growing incredibly fast. there are a couple of things i can say with absolute confidence. there will be more lines of cobol by the time we leave this room than when we stepped into it. [laughter] legacy businesses are great businesses, and they live for a long time. they live as a factor of human time, not as a matter of technology. technology can do the coolest stuff on the planet. is what humans do with technology that matters. so there will be more windows.net programs and all of that. the legacy will remain. what is driving public cloud today is predominantly consumer experience. that is growing incredibly fast.

of course, we are all getting new devices all the time. a fundamental change as these new devices, and is instead of having to maintain their own copy of windows the way they always used to, at&t is doing that for them now. it is brilliant. that is why they get out of owning it. because they do not have to service that thing on the device. consumerization is really helping i.t. and i think can really help move the whole computing experience forward, while preserving a way for people to access legacy applications that can be cloud- ified as well. >> i'm going to go right turn on the spirit let me start with something for you to think about. 80% of the u.s. economy is a service economy, right? we do not manufacture anything. we do not grow anything. we are a service economy. you probably heard this. probably every one of you in this room is in the service economy.

you probably are not growing or manufacturing anything. the question i would put in front of you is what the hell is a service business? what is a service economy? right? is it answering the phone nicely? is that a service business? is it flipping burgers burgersin-n0out? i get to say that around here, right? i would contend you that a service business is a business which delivers information that is personal and relevant to you. information that is personal and relevant to you. whether that means i go down to mark hopkins or wherever and talk to the concierge, and i really would like to find a cheap chinese restaurant that i can walk to that is less than $5, then he gives me the right answer or she gives me the right answer, or i go to my doctor's office, and this is based on your gino and lifestyle, we need to put you on lipitor and need to exercise more, right? that is service, right? let me take you to computers.

visualize the amazon website for a second. what are they trying to do there? they are not trying to deliver information that is personal to you, right? people like you bought this. people like you listen to this, right? i will ask a question. for the technical people in the room, where is the transaction processing system on that web page? i will tell you the answer -- it is that little shopping cart in the upper right-hand corner. most of us in the tech field -- i have been around for a while -- we have all worked on the shopping cart. we have worked on transactions, processing systems, they met. how important is it on the web site -- it has to work, and has to work in scale, and all that, but it is not very important, right? what is important is they are trying to deliver information that is personal to you.

let's take our favorite banking website. go to the website. you have to log in, right? from that point on, all you see is a transaction processing system, right? it does not actually care who you are. it has no idea who you are. the consumer internet sees " is 200 terabytes. that is all it is. not very big. the challenge is how do you get it. how do you take that information? i hate shopping at home depot,

okay? i do not really like going there. however, think about this scenario. what if three weeks ago, i bought tile, and two weeks ago, i bought a faucet, and last week, i bought a sink, and this week, i bought a vanity mirror? what am i doing? probably remodeling my bathroom, right? by the way, they also could know that i have actually spent $10,000 over the past four months at home depot. with all that information there, why isn't the managers showing up when i show up at home depot showing me the three toilets that would match my title, right? my only point is all that information is there. it is not being harnessed to deliver information that is personal and relevant to you. this applies to finance assistance, health care systems, city systems. it applies across the board. the major step that we will see

and people who innovate at this level will see is doing that. i will end by telling you a little trivial metric for you to walk away and think about. it is called servers per employee. all of you guys who work in a business, just to this for fun. go in and ask how many servers you have. big ones, little ones, i do not really care. abide by total number of employees, contractors, not contractors. a number in your head, some of you? i will give you a couple of data points. turns out the u.s. government tracks this for global economic development purposes. by the way, if the number of servers per employee is zero, what is the chances for global economic development? 0. right. and that brazil is a 0.04. india is a 0.02. i have no idea what the u.s. is that. that gives you some metric. my friends who run these application cloud service companies, they are sitting at

0.5 to 1.0. most of the people who listen to me -- actually, i did this three or four months ago. the guy sitting right here was the cio of kimberly-clark, an old friend of mine. so i asked him how many servers per employee he had, and he said we had 0.4 and headed for 0.2, proudly. ok, who do you think is sitting at 30 to one? facebook. would you think is sitting at 50 to one? for your technical people in the room, do you think they are operating inefficiently? no, i do not think so. furthermore, do you think that they are using all those servers to serve up web pages? no. right? these are truly information- power businesses. at the end of the day, my opinion -- lots of people may

share this -- is all companies and that being information companies. all companies and of being software companies. i just did he know speech for a large german car company that you may know, that is clearly when you start realizing what they are trying to do, right -- it is really can we use all this cloud technology to takeoff structures down and across this down so now, i can change and 8% spent on a bunch of stuff that other people can go due to 80% i get to build applications, which power a very different experience. you think about what the car of the future might look like, it is a computer with four wheels, right? that is what it is. what they start to do with it, etc., is completely different. i'm talking about what we all conceptualize as a manufacturer. when you ask the question what will happen in the future? it is harnessing this technology to really deliver a service economy, and the companies that do this, the guys that figure

this out are going to be big winners, and they are going to change the way we think of them, the way we relate to them, the way we buy from them, all of that. that is what the future holds. i see the floor. >> thank you. i think the best questions are yet to come, and we are going to turn it over to the audience. >> we would like to remind our listening and viewing audience that this is a program with the commonwealth club of california on the future of cloud computing. our thanks to our distinguished panel for their comments here today. now, we open the floor for a q&a session. we will be passing around a microphone, so if you have questions, please raise your hand and speak into the microphone. >> i have a key question about the backup plan. you mentioned the super bowl earlier. what is the backup plan in the unlikely catastrophic event of

the disabling of the system? solar storm or whatever. >> there are lots of things that can go wrong. the rights can hit the planet, and the things go dark, and then we fix it. in general, the technology you are talking about is something which is broadly called cloud bursting, where essentially, is used by google and other folks, there's not one computer called google.com. there's a basilian sitting behind a thing. there is a concept called load balancing, and it has been augmented of late with the ability to dynamically spinoff new instances of server applications in response to spikes in demand. the general concept called cloud bursting allows you to do that across multiple cloud vendors, so you could do it across amazon and various other people say you could get geographic diversity and so on. people doing this extremely well, for example, would be netflix.

many of you in this room i'm sure use the netflix. what they did is dynamically throw what is this is as more and more people click on movies that they want to watch. so then what they are doing is as the need scales, they then have the ability -- they pay them, and, of course, it drops off as soon as the need drops off as well. so they end up essentially paying for average demand. the technology is widely deployed around the world. >> as kind of a follow-up, individuals often use cloud services for backing up their computers. are we about to see people using their computers to back up their clout services to guarantee that they hold on to their data? >> you are at the tip of a very interesting iceberg. go 100,000 miles in space and

look down at the earth. we are still driving more wavelength down. there is no problem distributing content out to users. what we have a fundamental problem with is distribution of power. power is dominating in terms of distribution networks q one of the reasons why it does not make sense is because you are at the end of the tree, a long way from distribution, a lot of transmission loss. the data centers move to where power gets generated. what is the next hardest thing to move? big data. did it is still really hard to move, even though we have lots of wavelets. that says that the application moves to the data is. if you think about facebook and google and all those folks, they build data center's right were the power is generated. typically near hydro plants and so on. what you find is that more and more applications will move to where the data is. moving those big chunks of data is very difficult. in terms of enterprises in the cloud, there is certainly no

reason to suspect that the systems used by cloud vendors like amazon are not capable of geographic replication and redundancy. it is absolutely the case that someone like netflix, for example, could survive an outage of two simultaneous amazon did a centers through geographic redundancy and so on. this stuff exists, and the technology exists within the cloud providers to make sure that once your data gets there, it is not going away. it is just not a cloud provider solution. most enterprises will have things called disaster recovery, where they will keep all their critical data completely synchronized so that if one coast gets hit by an earthquake or terrorist attack, you have the other coast where you can get your data. it is completely synchronized, always available, on demand. geographically diverse disaster recovery solutions have been in

place for some time, and they actually do allow for secure data storage. >> i think that for the individual consumer and home user, this storage in the cloud and backing up your personal computer in the cloud has been burgeoning of late because people want to have the ability to store their files securely, but the reason why, to answer your question, is why do they continue to still backed up their clout back of solution onto a usb stick or on to their own pc is people still have to get comfortable with the idea of clout security, that the data truly is secure and they're comfortable with letting go. people still not quite comfortable with that concept yet. as people become more and more confident and more and more comfortable with the concept of data being safe, we will still

have people, and we will still have instances where people want their data next to them, where they feel comfortable, where they feel safe and confident that their data is secure. >> if i were purchasing services from a cloud vendor, i would mandate that all data at risk is encrypted using keys that i own, that i provide when it is processed for me, and there is no excuse for anybody not doing this. the technology exists. so it comes down to the probability that a bad guy could go and guess relocation. in amazon web services, you have more than 3.5 billion objects. they have to know which one to go for it here that have to break your access. then they have to break the description on the of jets. the probability of stealing your data and getting away with an attack is vanishingly small. >> i think you address one of my concerns, which was the security of my data out on the cloud, but you raised a new question -- who

owns my data? what if i do want it to go away? >> there are very challenging concerns. certainly, governed by state boundaries. for example, numerous canadians do not want their data in american dissenters. under a land the vessels, that can be subject to inspection and seizure. all the regulations relate to national boundaries there as well. a cloud providers actually end up having to meet numerous diverse regulatory requirements related to where data may resign and how it may be encrypted. there are different purchase centers for different christian center's and countries -- there are different encryption centers for different regions and countries. i think amazon wishes they could make it go away the same way they imagine the sales tax would go away sunday. [laughter] >> but is it clear