Amazon Adds Cloud Data Center in Virginia

As Amazon’s cloud continues to grow, the company is investing in real-world brick-and-mortar data centers to provide additional capacity. The retail/infrastructure company recently leased a 110,000 square foot property in northern Virginia to expand its data center footprint.

The additional space will help accommodate dramatic growth for Amazon Web Services, the suite of services that allow companies to run their applications on Amazon’s infrastructure and pay based on usage. More than 500,000 developers are now using AWS, and Amazon’s S3 storage now houses more than 50 billion objects.

Northern Virginia has always been a key market for Internet infrastructure. But the data center expansion may also reflect Amazon’s ambitions to host cloud applications for the federal government. Last week Amazon’s AWS Federal unit held training sessions for IT contractors who already have relationships with federal customers.

While other cloud builders like Facebook add computing capacity by leasing turn-key “wholesale” data center space to save time and money, Amazon is building out its own infrastructure in its new facility in northern Virginia. It’s not clear whether the timing of Amazon’s equipment purchases will allow the company to qualify for the newly-passed financial incentives for data centers in Virginia. The state recently enacted a new law providing a sales tax exemption for companies that buy or lease at least $150 million in computer equipment between July 1, 2010 and June 30, 2020 for use in a data center.

Amazon Pitches The Federal Cloud

Can cloud computing pioneer Amazon Web Services be a major player in the push to move the federal government’s infrastructure into the cloud? Or will newer cloud offerings from system integrators and managed hosting companies be the major beneficiaries of the Obama administration’s efforts to migrate government apps into the cloud?

Amazon is making a major pitch for federal business, as revealed today by Kevin Jackson at Cloud Musings, who describes training sessions in which Amazon Web Services (AWS) Federal is training IT services companies to leverage its cloud as the platform for outsourced government clouds. AWS Federal appears to be building its government pitch around contractors who already have relationships with federal customers, but may not possess the infrastructure to provide scalability for large cloud applications. Jackson writes that the training was “invite-only and attendance was IT services firms that had demonstrated a clear track record of success in the Federal market.”

Amazon’s interest in the federal market was first noted back in January by TechFlash, which spotted a job posting for a business development manager for “Amazon Government Solutions.”

Lightning Strike Triggers Amazon EC2 Outage

Some customers of Amazon’s EC2 cloud computing service were offline for more than four hours Wednesday night after an electrical storm damaged power equipment at one of the company’s data centers. The problems began at about 6:30 pm Pacific time, and most affected customers were back online by 11 p.m., according to Amazon’s status dashboard. The company said the outage was limited to customers in one of Amazon’s four availability zones in the U.S.

“A lightning storm caused damage to a single Power Distribution Unit (PDU) in a single Availability Zone, the company reported. “While most instances were unaffected, a set of racks does not currently have power, so the instances on those racks are down. We have technicians on site, and we are working to replace the affected PDU.”

EC2 previously experienced extended outages in February 2008 and October 2007.

Is AWS Elastic MapReduce A Game Changer?

Last week Amazon Web Services introduced the latest addition to its growing suite of cloud computing services: Elastic MapReduce, a hosted Hadoop framework that offers AWS users the ability to process huge amounts of data. This strikes me as an important offering for Amazon. Here’s what other cloud watchers are saying:

• VentureBeat says that Elastic MapReduce “marks Amazon’s movement away from just providing the bare-bones infrastructure and towards offering services on-top of the infrastructure.”
• Dana Gardner at ZDNet believes this could be a game-changer for Amazon. “Think of it as having your own tuned supercomputer that you can plug gigantic data sets into and ask questions that will determine the course of your businesses for the next decade,” he writes. “And you can pay for the pleasure on a credit card.”
• ReadWriteWeb says that “MapReduce is the new, big hammer, and as developers start looking around, every dataset starts looking like a nail.”
• James Urquhart contrasts Amazon’s hosted Hadoop service with Cloudera’s distribution, which is designed to speed Hadoop implementation in customer data centers or as a machine image AWS.  

Amazon: $86 Million in Servers in 2008

Amazon bought more than $86 million in servers from Rackable Systems in 2008, surpassing Microsoft as the server maker’s largest customer. The investment in data center hardware is likely tied to the growth of Amazon Web Services, the retailer’s fast-growing cloud computing operation. Amazon (AMZN) doesn’t break out any revenue data for its AWS operation, but recently noted that its S3 storage service now stores 40 billion objects, an increase of 11 billion from October and three times the volume from a year ago.   

Amazon spent $56 million on servers with Rackable in 2007, but boosted that to $86 million last year, as first noted at TechFlash. The growth of Amazon’s business with Rackable has bucked a trend in which the vendor’s other marquee customers, Microsoft and Yahoo, have reduced their spending.  

Roundup: Amazon Reserved Instances

There’s been some interesting analysis and commentary on Amazon Web Services’ announcement of Reserved Instances, a new pricing model for its EC2 compute-on-demand service offering customers the ability to reserve large amounts of capacity for future use. Here’s a roundup:

• Will Amazon’s capacity be there when you need it? This is a key question, since Amazon is touting the feature for disaster recovery. Craig Balding at Cloud Security looked at the fine print in the AWS Reserved Instances agreement, and found that Amazon says it “may terminate the Reserved Instance Pricing program at any time.” “If you want to use AWS for Disaster Recovery, you better have a plan B as Amazon will only give you 60 days notice if they decide to drop Reserved Instances,” Craig writes.
• Reuven Cohen at Elastic Vapor believes Amazon is using Reserved Instances to target a new market. “They are going after the web hosting space,” Reuven writes. The new pricing may offer easier comparisons with traditional web hosting plans, but the missing link is e-mail services, which are an essential component of most small business hosting plans. I don’t believe it’s currently feasible to run e-mail servers atop AWS, although spammers have tried, leading many hosts to block email from AWS. One workaround would be to split off e-mail services onto Gmail, a strategy being encouraged by some large hosting providers. The hosting ecosystem relies heavily upon resellers and affiliate referrals to route business to major players. But it remains to be seen whether many hosting resellers have the technical skills to operate atop AWS, and although Amazon has an active affiliate program, it seems unlikely to match the $100 referral fees offered by major shared hosting companies.
• Amazon EC2 may not be a feature-for-feature replacement for web hosting accounts, but its pricing model is evolving to enable better ROI comparisons with hosting companies and data center providers. Geva Perry has put together an Amazon Reserved Instances Savings Calculator that looks at some of the math.

Amazon S3 Now Stores 40 Billion Objects

Amazon S3 (Simple Storage Service) is now storing more than 40 billion objects, an increase of 11 billion from October and three times the volume the service stored a year ago. The new total was announced yesterday at the USENIX conference by S3 General Manager Alyssa Henry, whose PDF presentation is available online (link via James Hamilton).  

The user base for Amazon Web Services is also growing fast. AWS said yesterday that it now has 490,000 developers using its platform, up from 370,000 last April. The new data quantifies the rapid growth of Amazon’s utility computing operation, and suggests that the company’s underlying infrastructure is likely growing fast as well. Check out our Amazon Channel and Cloud Computing Channel for more coverage of the Amazon’s cloud computing operation.

Amazon: Coy About Containers

Amazon (AMZN) likes to talk about its cloud computing operation, but usually has far less to say about about the data centers that support it. The company has been secretive about the facilities supporting its fast-growing Amazon Web Services operation, even when it added “availability zones” offering geographic redundancy. Amazon technology evangelist Jeff Barr once said he “has no clue where they are … they’re carefully protected.” It was only upon the launch of the CloudFront content delivery network that we learned any of the locations of Amazon’s data centers.

Amazon senior vice president of web services Andrew Jassy did an interview with TechFlash this week in which he talked up enterprise adoption of cloud computing, but wouldn’t say how many people work in Amazon Web Services. TechFlash’s Eric Engelman used the recent arrival of former Microsoft researcher James Hamilton as an opportunity to ask about data center containers. An excerpt:

Q: Will Amazon be looking more seriously at the containerized model?

Jassy: We’re going to look at a number of different — there a lot of different ways you can go with data centers and that will be one of several approaches we will look at.

Q: Does Amazon use containerized data centers now?

Jassy: We don’t disclose that.

Does Jassy’s inability to simply say “no, Amazon doesn’t use containerized data centers” mean anything? Only continuing intrigue about Amazon’s infrastructure.