Cloud computing has evolved from the realm of novelty to a major growth industry in just a few years’ time. Enterprise-level companies and startups both rely on this computing infrastructure to increase efficiency, scale more easily, and lower costs.
But how will the Cloud change in five years – or even six months? How will current Cloud providers adapt and how will they deliver for a rapidly growing customer base?
Tech evangelist Robert Scoble of Rackspace (and of popular blog Scobleizer) will answer these and other questions about the Cloud during his panel at Web 2.0 Expo New York this October. (Also on the panel: Lew Moorman, Chief Strategy Officer of Rackspace; Allan Leinwand, Chief Technology Officer of Zynga; Ari Levy, tech reporter for Bloomberg News; Brad Hargreaves, co-founder of General Assembly, a New York startup incubator.)
Scoble recently spoke to us about what they’ll cover during this session as well as his own views on where the Cloud is going. Read the interview, which has been edited for length and clarity, below.
Kaitlin Pike: In your session title you have this juxtaposition – asking if it’s a bubble or a revolution. Why can’t it be both? If it’s revolutionary, it’s sort of natural that it becomes a bubble.
Robert Scoble: Absolutely. And it certainly does get over-hyped because so many companies have adopted it. The bubble part is also the industry is heating up and we’re seeing that in valuations … That’s what people are usually thinking of when they think of bubble… People use the word Cloud to mean so many things that the word has become almost completely useless. In fact, you had to ask me how I’m using the word, just to make sure you were clear on that.
Pike: So because of this, does it matter how many players there are in the space?
Scoble: I don’t think so although, you know, when I go around and interview startups, I only hear two names, [Rackspace] and Amazon, for the most part. In enterprises, you hear more of the infrastructure players like VMware or Citrix, or Microsoft. And then there’s a bunch of lower players like GoGrid and so forth. Generally people are coalescing around a few names.
Pike: Many companies who might want to move to third party Cloud services simply can’t because of security and privacy concerns, the control, support, or even regulation issues. How do you address these concerns – particularly security and regulatory questions – before companies, you know, from finance and health care can use a third party Cloud system?
Scoble: Well, you can address it a number of different ways. At Rackspace, we have traditional hosting as well as Cloud, so we can segment the kinds of services that are mission critical and that you feel can’t be shared with anybody else, but you really have some security concerns or whatnot. Usually people are okay with that being in a third party data center. They just might not want those resources shared because that gets into a new kind of risk. If it’s talking about Cloud services, then we can get into the whole discussion of how protected are these services from bad neighbors, how protected are they from hackers, how protected are they from a bad actor at a host company? … It will be interesting to see what Lew says on all three of those and how he takes on some of the fears that existing enterprises have.
The other ways… you can keep doing your own data center but move to Cloud technologies internally so that you get the energy efficiency and other cost savings that comes from sharing resources internally. And that’s when we will talk about OpenStack for instance and the advantages of going with that technology – open source technologies – inside your own data centers. It’ll be interesting to see the panel talk about that – what other choices that they have for internal data center systems.
Pike: That’s actually a great segue to one of the other questions I had, talking about making it more open source. One current problem is that once you use a service, you tend to be locked into it. So for instance, if I wanted to use Rackspace and Amazon, it’s not exactly easy. What are different services doing, obviously including Rackspace, to create standards that allow me to use multiple services?
Scoble: Amazon is trying to create a de facto standard just because everybody is using it. And Rackspace is trying to compete with that by doing the OpenStack and the open source OpenStack. The OpenStack advantage is it’s completely open source; you can run it on your own servers. You can’t run Amazon’s infrastructure on your own servers. So there is a pretty big divergence between those two systems. One is an open standard that you can actually see the code underneath. One is a closed standard, which is the de facto style standard. Are there players who are trying to get standards like that? I don’t think there’s that many that have that kind of market power. Google is trying to do its own big table sales standard. I very rarely hear about Google or Microsoft in this world. I almost always hear Amazon or Rackspace as the choice that people are making, or OpenStack and Amazon.
Pike: Do you think it would be smart to have someone like Google or a different organization work on these standards because I mean it’s, I don’t mean this to sound insulting, but it’s sort of a conflict of interest for Rackspace and Amazon to be working on standards because it’s part of your interest to keep people within your own system.
Scoble: Google has the same interest.
Pike: That’s true.
Scoble: And so does Microsoft, you know. At Rackspace we welcome… anybody who’s interested in an open source standard to sit down at the table. And I think that’s pretty open. And we have NASA, we have Dell, we have at least fifty different companies now at the table. So, yeah. I don’t think we’re saying “stay away” to Microsoft or Google. I think they would be welcomed in this kind of effort, and certainly they would be welcome in proposing different kinds of APIs that they feel their customers would need.
Pike: Okay. And then, from these things you need to standardize, what do you think will get standardized if it hasn’t been already, and what do you think will never be standardized?
Scoble: That’s an interesting question. It seems like the file sharing, the databases, the standard APIs are fairly open at this point. The stuff that I think is going to be tougher to standardize is billing systems. Rackspace has its own billing system that’s worked on or some support network. We’re not going to share that probably with the rest of the world and that’s going to be how different companies differentiate. You know, Rackspace made a very strong choice to give away the infrastructure but compete on service and that’s what people pay us for: The ability to call at 2 in the morning and get a human before it rings three times and that’s where the proprietary stuff comes in.
I mean, if you look at the latest investment we made in Cloudkick, it’s management systems also. You know, being able to keep these systems up and running and make them accessible to customers in a different way. Those might be places where we stay a little more proprietary, you know, while the whole cloud stack is actually open sourced. But it’s pretty easy, you know, to look at our code and see where everything is open. And then you can say, “Okay the billing systems aren’t here so we need to build our own billing system. We need to build our own support network if we’re going to offer support. And we’re going to need to build in maybe analytics.” And it will be interesting to see what Lew says to this question and we’ll see where the conversation goes.
Pike: Do you have any idea of where it’s going to be, say, five years from now? Do you have any expectations for what the standards would be like?
Scoble: I don’t. That would be a question over my pay grade. Yeah, that would be a good one for Lew. And certainly I think the people who are building, you know, bleeding edge infrastructure, the guys at Zynga, would probably have a better idea of where the cloud is moving because they’re having to manage, you know, many, many thousands of servers and huge loads and they’re having to innovate and come out with new things that will help them keep on top of their business and save costs and they’re going to be the ones who are going to come out with some interesting new things over the next few years. I’m sure Lew can tell you about what Rackspace is investing in and where we see the customer needs. But I think Zynga’s going to be the one to really talk about their needs, because they’re hitting walls that most of our customers are not yet hitting.
Pike: What is wrong with cloud computing today, what do you want to see changed?
Scoble: I’d like to see it become more programmatic. I think that’s something that I keep hearing. For instance if you get on, if you’re building a business for let’s say Major League Baseball. And then the World Series comes on, you’re going to see a spike in traffic and you don’t really know how big that spike is. So you have to either have to have people sitting, watching systems and spinning up servers and moving things around or you can have a system that programmatically opens up servers and then when the traffic dies down the next day it starts killing servers and starts reconsolidating. It would be interesting to see how that innovates. Again, that’s a question really for Zynga to talk about because they’re hitting that kind of spike problem much more than the average startup is. I mean the day after Christmas when everyone gets their new iPads, the first thing they do is start playing games and they see huge spikes and they have to deal with that, and it will be interesting to see what kinds of problems they’re hitting.
Everybody always wants lower cost. So, how can we save electricity, how can we save on server space, how can we make things more efficient and cheaper? We are always interested in that, that’s why I loved visiting SeaMicro, which is making these really small, very energy efficient servers. That kind of innovation will allow new kinds of cloud infrastructure to be built for lower cost and less utilization of electricity and other resources.
Pike: So, I have question to follow up with from that. How “green” is the Cloud? It’s obviously a concern that we have.
Scoble: It depends how you look at it. On one level if you just look at how much electricity these data centers now use it’s a huge amount, and it’s growing at a huge rate. Both us and Amazon are seeing growth that’s astounding and so we are faced with demand, customer demand, that is forcing us to buy more data center space and build out and use more electricity and more heat, more cooling systems, stuff like that. On the other hand, if you look at it the other way which is comparing an old data center inside some company or a few servers in a closet, inside a work group or something like that. Those servers are only being used – I believe the industry averages around 15 to 20 percent – so if you move those workloads to Cloud computing, you see a huge environmental impact because you’re sharing those resources with other projects and other companies and your utilization on the server goes from 15 to 20 percent to 80 percent and that’s great for everybody.
And it also means building these massive data centers. You saw my tour of the Facebook data center. Well that’s a small one compared to the data centers that are down in Las Vegas and elsewhere. Centralization of everything makes them more efficient and makes it possible to upgrade equipment to something like what SeaMicro was doing with microservers that used a quarter of the electricity with four times the compute power in the same rack space.
So that’s very exciting and that’s going to bring the whole industry as a whole more energy efficiency. But this industry is growing like a weed, and it’s demanding a lot of energy, and that is a big problem for us.
Pike: What do you think is the next evolutionary step in the cloud? Is it just interoperability and address security, that sort of thing?
Scoble: I was just at the VMware’s announcement and their evolutions are hundreds of little things that are memory management, that are memory utilization, newer architectures that take advantage of SSD instead of using hard drives. So that’s allowing the guys who build these virtual machines to make optimizations that weren’t possible before: New kinds of database architectures. We invested a lot in the MySQL or MySQL Drizzle, which is a lighter weight database engine, optimized for these internet workloads, these Cloud based workloads, and that makes everything more efficient and makes systems – scalable systems – possible. Does that answer that?
Pike: That definitely answers that.
Scoble: So there’s a good infrastructure on the hardware innovation that’s going on, like the SeaMicros, there’s good innovation in database architectures like my SQL Drizzle and then the Cloud stacks getting more efficient. For instance, as NASA started working on OpenStack they found optimizations that they could put into the code and [then you all can see] optimizations and the whole stack is evolving very quickly on all three fronts. It’s pretty exciting.
Pike: What have you always expected someone to do with the Cloud that you haven’t seen yet?
Scoble: I don’t know that I would have expectations. I certainly see where the bleeding edge is, you know, the Twitters. Twitter is only four or five years old, and it just this past year got robust in dealing with all these messages, and you can see there’s billions of tweets being generated every day. I’m following 32,000 people, and I see the tweets just streaming down my screen like the Matrix. We haven’t yet seen the end of that, because, you know, Google just came out with Google Plus which has very good circle management. But that’s just the start. We need to be able to see trends in those circles and we need to be able to say, “Which items have been shared the most here?”
We haven’t even gotten to that point. So I would say the stuff that I hope to see the Cloud move toward is much better real time information about what is going on on our screens.
Questions from the Web 2.0 Expo Audience
Pike: Okay. I have a couple questions that I got from Twitter and Facebook when I mentioned that I was going to be interviewing you. So, first question from Twitter: What is your Google circle strategy?
Scoble: Survival. For right now, I’m just watching the usage pattern that people are putting in there. For instance, I started a search for venture capitalists. And the venture capitalists are not talking about anything I’d expect them to. They’re not talking about funding or economics or term sheets or strategies of dealing with entrepreneurs or anything like that. They’re talking about all their fun vacations. And Joi Ito is showing his scuba diving photos and other people are posting fun photos and stuff like that.
So my strategy right now is to keep the number of circles down to smallest possible. Don’t go crazy because your assumptions right now are probably going to change over the next month. And just keep watching people and then watch how the usage pattern evolves. … I’ve already added a circle for photographers because I’m noticing some really wonderful photographers on Google Plus.
I’m just watching what happens and how it’s different from Twitter and Facebook. I’m trying to build circles that accurately reflect the usage that’s going on.
Pike: Next twitter question: Can you talk about the Scoble Effect? This person is referring to the recent TechCrunch article talking about celebrities quickly dominating conversations on new social networks.
Scoble: Yep, yeah cause I move in and my crowd does too. And we play with it, and we all react to each other. I think that’s an effective network system that we’re part of.
If you’re a popular person in society or in this case, early adopter geekdom, you can cause funny things to happen, you know, when the system is new. Usually people call it the Scoble Effect when I drive a lot of traffic to their company and their servers go down or something like that. It’s more like the modern Slashdot Effect.
And that’s sort of funny, but it’s not really true. What’s going on is, if I only have, let’s say I have forty followers, and I say something really outrageous, those forty followers will push my message out to their followers and so on and so forth and if something is really interesting that I say, by the end of the day it hits a million people.
That’s not necessarily due to the network that I have following me, but it’s due to the fact that I’m pretty good at causing conversations to happen and those conversations spread to go viral. It’s not something I can predict. For instance I was one of the first people that talked about Flipboard, and everybody saw that post tried it out and said, “Yes, this is right. This is awesome.”And they kept pushing it to their followers, and yeah, within a few hours, their servers were going down. I’ve tried that with other companies that I thought were cool and everyone’s like, “Nah Scoble, you missed the boat this time.” And, you know, they only got a couple thousand visitors.
I can’t always predict when things are going to really resonate with people and when they’re not. But I do have a lot of followers. I can’t deny that anymore. And those followers tend to be people who are very early adopter, very passionate about technology, very passionate about trying things. And so that causes a big bang effect that everybody notices, and sometimes it pisses people off.
Pike: Speaking of that, pissing people off, other question I have is “When are you going to retire?”
Scoble: When I’m dead. I mean, I’ve done this before I got paid. This is my life. When I was in junior high I got a tour… I walked up to the front door of Apple Computer when it was one building and said, “Give me a tour.”And I was that way when I was in junior high and I’ll be that way when I’m 80, you know, if I can still walk.
I don’t understand that concept of retiring. I mean, I would like to take some more trips as I get older and slow down a bit. But I will still be sitting around the pool going, “Hey, did you read what happened on TechCrunch today?” or whatever it is.
Pike: So this next question I think is my favorite. It started off as a joke, but then we Googled it and it turned out that tons of people are asking this question. And it is “Are you a Cylon?”
No, you know, I have biased my entire life to being public and causing conversation, so my behavior looks robotic-like at times. Ask my wife that question, and it’ll get you a different answer.
About the author: Kaitlin Pike is the Marketing & Community Manager of Web 2.0 Expo. She can be reached @kcpike.