Cloud services are booming, and Akamai is looking to capitalize on that growth by providing differentiated services based on its distributed network architecture. While it does offer optimization for cloud-based services, CEO Paul Sagan said that the provider would not seek to compete against Amazon and Rackspace to provide the same sort of bare metal, infrastructure-as-a-service cloud offerings that they sell.
Akamai CEO Paul Sagan will be speaking at this week’s Structure conference. In a Q&A ahead of the conference, we asked Sagan about how Akamai defines its cloud services, why cloud optimization is important and how companies that provide cloud services can overcome questions about security or network outages. To learn more, and to see Sagan’s fireside chat with Om on Thursday, join us for Structure 2010, June 23 and 24 in San Francisco.
NewTeeVee: A few years ago, Akamai tried changing the language with which it describes itself — not as a CDN, but as a cloud services provider. What’s the difference?
Paul Sagan: Well, I think the language of the industry has changed and evolved. If you really think about some of the basic tenets of cloud, Akamai has operated as a cloud provider since we began 12 years ago. Content goes into the network and people don’t know really where it is, but people connect and get transformation done on the fly, get access to content, get access to applications, get results back on their screen. That’s really classic cloud, as we would think of it today, but if we described it that way and called it a cloud 12 years ago, people would have just scratched their heads.
So in the last few years, as the concept of cloud and the vocabulary of cloud has shaken itself out, we started talking about Akamai as being a cloud operator. Most of our services in this area are around optimization of uses of cloud infrastructure. Most of what we do in that space is actually optimize infrastructure as a service for people, including those who are using other cloud infrastructure.
NewTeeVee: Can you give an example of how a customer might use you for this sort of cloud optimization?
Paul Sagan: There are three classic layers on the cloud: there’s infrastructure as a service, application platforms as a service and software as a service. We offer optimization in each one.
In the infrastructure as a service, I’ll give you a concrete example. There’s SmugMug, the photo sharing site. They’re actually using AWS [Amazon Web Services] for hosting, but they use Akamai’s optimization services to make their applications and delivery work at levels that satisfy the end user. So they get about a 300 percent improvement in performance.
Force.com uses Akamai’s optimization services for the more than 100,000 applications that run on [its platform]. They’re getting better performance and big offload from the centralized infrastructure where they’re hosting these applications. That’s platform as a service — it’s above the level of infrastructure and renting the bare metal.
Then at the software layer, we have over a hundred different software-as-a-service providers use us. SaaS — even if it’s replicated in a couple places — [is] effectively centrally hosted software. It’s accessed globally, often by a very large proportion of its users outside North America, and these applications often don’t perform well to the end user without Akamai. So there are over a hundred providers who use us, because we guarantee and ensure the right performance to the end user’s screen, even if it’s hosted in a big data center far away.
NewTeeVee: Do you have any interest in offering the same sort of services as Amazon and Rackspace?
Sagan: No, that’s not what we’re going to provide. We want to provide performance optimization on top of that layer. The only caveat is that we do sell storage in the cloud, but it’s not generic — it’s very specific to our media customers and is optimized for the delivery of media. We don’t plan to set it up so that you can come along with your credit card and say, “Can I put five files there, and you’ll bill me a dollar?” It’s very specifically tied to our media, because there we find you need highly specialized infrastructure.
NewTeeVee: When customers are looking at do-it-yourself versus moving to the cloud, what are the things they’re concerned about?
Sagan: The first thing that they really think about is security. The worry is, “I’m letting go, is the offer secure?” I won’t speak for all cloud providers, but our pitch is: We focus on security with far more resources than you will. Our denial of service protection is filtering traffic far far away so that a denial of service traffic never even gets to your data center. If an attack has made it to your data center, it’s too late already. So we believe not just in the cloud, but in a decentralized infrastructure. We can provide greater security from a lot of the threats that are out there, stopping denial of service attacks, filtering malware and trojans and viruses, et cetera.
I do [also] think we can provide better uptime through the cloud and a distributed model. For example, we target four nines of performance for our services or better. The pipe alone will never give you that performance to all your end users, and the data center will rarely give it, because of the single point of failure. So going to a distributed and a shared outsourced model can get you closer to the level of ability that people are actually targeting.
NewTeeVee: But there have been some high-profile outages of cloud services over the years. How do you fight the perception that these services are not robust enough for enterprise use?
Sagan: Not everyone’s going to do a good job, and customers should expect that their provider is going to meet whatever SLA [service level agreements] they paid for. I do think there will be companies who have better practices than others, but…we’ve all experienced different kinds of outages or failures or security breaches, and some of them get lots of publicity, and some of those were within corporations: credit card information getting out, email being down, software being buggy and crashing, applications that failed.
You’ve had the experience of being in the workplace and some location stopped functioning, right? So it’s not that it’s perfect and now moving to bad. I think what every company has to do is weigh how well they can do on their own, and at what cost, versus how well their cloud provider can do, again at what cost.
There will be some companies that provide very, very high levels of uptime — that’s Akamai’s goal — at a reasonable cost. And there will be some providers who will say they can do it at that level and they’ll fail, and they probably won’t succeed very long. And there will be other companies who will say, “Certain things you want to do, it’ll be good enough, but it will be a lot cheaper.” That won’t work for a financial services or a medical application, but it might for some personal use for backup or something. Not that you could lose it, but it might be a few minutes delayed. The use case is very important: If you’re doing online trading, fractions of a second can matter. But if you’re trying to restore a file, if you’re delayed a few minutes, it’s probably not quite as serious.
For more on cloud computing, join the GigaOM Network for its annual Structure conference June 23 & 24 in San Francisco.
Related content on GigaOM Pro: Report: Delivering Content in the Cloud (subscription required)