Q&A: Amazon on Enterprise Computing in the CloudQ&A: Amazon on Enterprise Computing in the Cloud

Among the earliest proponents of cloud computing, Amazon is also the market's pioneering provider, with services including Amazon Simple Storage Service (S3), Amazon SimpleDB, and the Amazon Elastic Compute Cloud (EC2). Speaking at this week's Enterprise 2.0 Conference in Boston, Adam Selipsky, VP of Product Management, Amazon Web Services, asserts that enterprise IT is already being drawn into the cloud.

You're on a panel at the Enterprise 2.0 conference in which they'll pose the question, "have we reached a tipping point where all of IT, including custom applications, can be moved into the cloud?" What's your answer?

I'm not actually going to say that I think all IT needs can and will run on the Cloud. We've built our services to be extremely flexible and open, because there's incredible heterogeneity out there, but it would be pretty preposterous to claim that we can cover 100 percent of computing needs. However, we do think that a huge swath of applications and a huge swath of developers can and will move into the Cloud.

Why is cloud computing attractive?

There are at least three big reasons. First, if you consider storage, compute capacity, databases, payment processing and everything needed to run a business, it all requires a bunch of heavy lifting. It can stop your business if it breaks, yet it really doesn't bring any value to your customers. Nobody cares where a photo management site stores photos, but without that capability, the application breaks. So it's a lot of work, and a lot of exposure, yet there is very little payback. Not having to worry about that muck — being able to focus on product development, sales, figuring out your strategy, and so on — is a much, much better use of time for most businesses. So that's one big reason why there's so much IT activity trying to move into the cloud.

Another reason is pure economics. [Amazon] is running a set of services that are large, rapidly scaling out and that are balancing the peaks and valleys of many, many customers with very significant [compute] volumes. We have very high asset utilization and, therefore, relatively low cost per unit of hardware, per unit of bandwidth and so on. We can pass the savings on to customers in the form of low prices, whereas, if you're sitting on a bunch of hardware that you're using at 10-percent or 15-percent utilization, which is normal for most enterprises, then you're not being very cost effective.

A third argument for cloud computing is performance. We've been doing this for Amazon.com for more than a dozen years. We've had a lot of successes, we've made some mistakes and we've learned a lot over the years about what it takes to run reliable, fault-tolerant, scalable infrastructure services. There are not that many companies out there that not only have the capability to build services like this, but that have actually had to run them.

I want to get back to reliability, but first can you describe those "big swaths" of would-be cloud computing users, particularly among midsize and large enterprises?

There are just so many applications that are good fit, but I'll give you a few general examples. To start with, anything that operates on the Internet and that is customer facing is incredibly low-hanging fruit... Second, we're seeing great uptake for big data-processing applications, such as scientific computing, drug company clinical trials and hedge fund mathematical analysis. These are crunch-intensive applications where you need large amounts of compute capacity, but you mainly need it sporadically.

Very generically, anything that's a spiky application is incredibly compelling to put in the Cloud because you can tap into, say, 500 machines to handle data processing for a week and then shut it down until you need it again. It's just incredibly efficient to be able to do that by just paying by the hour.

A Gartner analyst Intelligent Enterprise recently interviewed predicted it would take two to four years for the cloud to take root in the enterprise due to security and reliability concerns. Do you agree?

There is a lot of nervousness around security. It tends to be an emotional reaction, but it dissipates as we talk through those concerns. "Gosh, I'm worried about security." "Well, what specifically are you worried about?" "Well, I don't own the machines and the data is not within my four walls." "Well, do you have co-location providers?" "Yes." "So isn't your data not within your four walls there?" "Well, yes, that's true..." Software as a service is another example where the data doesn't reside within your four walls, yet SaaS is a category that is supposedly mature.

The question is not "where is the data;" the question is really, "how is the data treated and what are the protocols around it?" Amazon has a pretty good body of experience in dealing with confidential data like tens of millions of credit cards. We take that very seriously.

Part of maturation is an education process, and it will take time because human beings who are trying to do their jobs need to get comfortable with a very new paradigm.

Is it just a matter of getting comfortable, or does cloud computing need to evolve?

I think there will be some of each. Amazon has been on the market with Web services for a little bit over two years — which is two years longer than pretty much everybody else — yet we still feel like we have a lot of ground to cover. We're still releasing new services. We're still adding important features to services that we've already released. We're far from done with feature development, and that's driven by all the things our customers are asking for.

Gartner contrasts SaaS vendors with cloud vendors saying the former generally provide a lot more detail about where the data is stored and how it's backed up.

We will find ways to provide customers with more information about the resources they're consuming, but you have to do it in a way that's accurate and that also doesn't create security issues. You have to balance transparency and security.

I'd also say that part of the whole point of computing in the cloud is that you don't have to worry about where the resources are. If you decide to do enterprise backup in S3 — which is a great application that we're seeing more and more of, by the way — do you really need to know exactly where and when those copies are replicated and how our replication works? That's time you could be spending on something else. What you want to know is that we're never going to lose the data, that your people are going to be able to access the data whenever they want, and that it's secure.

What assurances can you provide on reliability? EC2 is in beta, currently, but you've had a number of outages [and last week, Amazon.com itself had a two-hour site outage]. What can you say about service level agreements (SLAs)?

The place where customers were most eager for us to start with SLAs was S3, so we did that. It's an availability SLA, and we actually put the equation of the error codes that go in the numerator and the requests to the system that goes in the denominator. If we fall below 99.9 percent availability, you get service credits. We don't have a separate durability SLA for S3 because we just don't lose your data; 100 percent is the SLA.

What about EC2 once it's out of beta?

When we make that service generally available, I would anticipate that we will put an SLA in place, and there are lots of possibilities. Is it about the API being available? Is it about the physical network infrastructure being available? We've been talking to customers about what they find most compelling and most reassuring.

Google and Salesforce.com are at Enterprise 2.0 to talk about their cloud offerings, and companies like Microsoft and IBM are also getting into the market. How will Amazon differentiate itself once there's more competition?

It can't be a good market without good competitors, but we think this is a big enough and interesting enough market that there will be several competitors that will do well. Undoubtedly, each will offer a different approach, but I can't speculate on what those others might do.

What I can say is that our strategy has been to have a series of building-block services that together form a compelling and cohesive platform so that you can outsource the infrastructure layer of your computing. That platform allows total flexibility for developers. We want Ruby, Perl, Java, C++ programmers to all be able to work with us equally well. We want Web developers and enterprise developers to be able to work with us equally well. We will continue to add things that make it easier and easier to use our services, such as tools and libraries and sample applications that let people consume these services in more of a plug-and-play way.

Would you say that Internet-based, customer-facing businesses like Amazon are at the center of the target and that enterprises business might be ceded, one day, to cloud vendors that specialize in enterprise services?

I would absolutely not say that. We had Fortune 500 companies using Amazon Web Services literally from the day we launched S3. It's true that the majority of our early usage was from smaller companies and startups because they tend to be a little more risk tolerant. We thought it would take several years to get to the enterprise stage in a meaningful way, but we've actually been rather surprised at how quickly enterprise adoption and interest have accelerated.

Tech Crunch published a stat that more than 60,000 corporate customers are using Amazon Web Services. Can you confirm that statistic?

I have no idea where they got that stat and I have no comment on it, but I can say that we are having more and more conversations and we have a multitude of existing, deployed relationships with large and midsized enterprises. That trend is dramatically accelerating and it's only going to continue to accelerate.