Where Amazon’s Data Centers Are Located

Amazon (AMZN) has announced the launch of its content delivery network, which it has dubbed CloudFront. The new offering, part of Amazon’s utility computing operation, offers content delivery with “low latency, high data transfer speeds, and no commitments.”

As part of the rollout, Amazon has also disclosed the locations where it will cache web content, offering the first public details about where its data centers are located. Here’s the list:

• Ashburn, Virginia
• Dallas/Fort Worth
• Los Angeles
• Miami
• Newark, New Jersey
• Palo Alto, California
• Seattle
• St. Louis
• Amsterdam
• Dublin
• Frankfurt
• London
• Hong Kong
• Tokyo

The locations of Amazon’s data centers has been a topic of interest since the company first offered geographic diversity for its AWS services last year. At the time, Amazon sad it could spread user content across “zones” in the US and Europe to improve redundancy, but didn’t address the locations of the content.

Amazon Building Large Data Center in Oregon

Amazon.com appears to be the tenant in a large data center rising on the banks of the Columbia River in Oregon, joining Google in harnessing the region’s cheap energy resources to power huge cloud computing data centers.

The $100 million data center is being built in Boardman, Oregon in the Port of Morrow, a 9,000 acre industrial park. Plans call for three large buildings on the site, according to The Oregonian, which reports that representatives of Amazon have attended local meetings to discuss permits for the site. The first building is underway and will be 116,000 square feet. 

Amazon is historically tight-lipped about the location of its data centers. But the rapid growth of its Amazon Web Services (AWS) cloud computing platform long ago exceeded the excess capacity in the data centers supporting the company’s retail operations. As AWS grows, Amazon will need to continue adding dedicated infrastructure to add capacity. Amazon recently said its S3 cloud storage service was now storing 29 billion objects.

Amazon Cites Momentum as EC2 Exits Beta

Amazon’s EC2 compute-on-demand service moved out of beta and into production today, with the key difference being that there’s now a Service Level Agreement (SLA) ensuring customer credits should EC2’s uptime fall below 99,95 percent. Amazon previously offered an SLA for its S3 storage service, but not EC2. Windows Server and Microsoft SQL Server are now available in beta for EC2, which is also adding a management console, load balancing and monitoring services.

These additions are the latest advances in Amazon Web Service’s transition from a playground for developers into a cloud platform offering on-demand services suitable for startups and enterprises alike. While the definition of “beta” has become decidedly fuzzy (Google has half its products in beta, including Gmail), there’s no question that beta status and the lack of an SLA are a barrier to adoption for many enterprises. EC2 has now eliminated those potential resistance points.

Amazon S3 Storing 29 Billion Objects

Jeff Barr from Amazon Web Service reports that Amazon S3 (Simple Storage Service) is now storing more than 29 billion, an increase of 7 billion from the previous quarter. “As one of the S3 engineers told me last week, that’s over 4 objects for every person now on Earth,” Jeff writes. At peak usage, S3 is handling more than 70,000 requests per second.

So what’s next? Amazon is lowering prices on S3 storage, with a new four-tier pricing plan that takes effect on Nov. 1. Customers storing more than 500 terabytes will get a rate of 12 cents per gigabyte.

When Amazon S3 was launched in March 2006, we wondered whether it would be a disruptive force or non-event. “It’s too early to say whether S3 (and the similar services that will certainly follow) is the start of something big or an experiment,” I wrote at the time. With 29 billion storage objects, it’s definitely something big.

Amazon Now Does Windows

Amazon says its EC2 utility computing service now supports Microsoft Windows Server and Microsoft SQL Server. This represents a significant expansion and opens Amazon’s platform to a wider circle of potential users. Here’s a roundup of the news and analysis about today’s announcement:

• Stacey at GigaOM raises key questions about licensing: “I wonder how the pricing model for using the Windows EC2 will look. Amazon has to license the software from Microsoft, so pricing may be higher. It’s also possible that Amazon could eat the cost, or that it worked out a favorable licensing deal. Since Amazon declined to talk about pricing yet, we’ll just have to wait and see.”
• “Even as it links up with Amazon Web Services, Microsoft is preparing to muscle onto its turf,” writes Nick Carr, noting that Microsoft will unveil its own “cloud operating system” next month (more on that from The Register).
• Bob Warfield at Smoothspan thinks the ability to run Windows and Linux is a big deal for Amazon’s cloud offering. “This is classic strategy because they’re making it tremendously easier for customers to pick one cloud computing (platform) and not have to choose … Amazon’s move really creates a more level playing field and is one more reminder to Microsoft it can’t afford to be isolationist forever.”
• Mary Jo Foley also sees this as a “pre-emptive strike” ahead of Microsoft’s cloud operating system.
• For Amazon’s perspective, check out blog posts from CTO Werner Vogels and AWS evangelist Jeff Barr.

Amazon to Launch Content Delivery Network

Amazon Web Services is planning to launch a content delivery network, adding a name-brand low cost competitor to an already crowded sector. Amazon’s entry figures to be bad news for the many venture-backed CDN startups seeking to challenge market leaders Akamai (AKAM) and Limelight (LLNW).

Amazon said the new service “will provide a high performance method of distributing content to end users, giving your customers low latency and high data transfer rates when they access your objects.” There will be no minimum fees or commitments - a common practice for some CDNs - and users will pay only for what they use. The service will deliver files stored on Amazon’s S3 storage service.

“This is an important first step in expanding the cloud to give developers even more control over how their applications and their data are served by the cloud,” said Amazon CTO Werner Vogels, who said the service is in private beta but will be “widely available before the end of the year.”

CloudBursting: A Hybrid Approach to the Cloud

How can businesses leverage the cloud without losing the comfort and control of in-house data center operations? Amazon Web Services evangelist Jeff Barr says many companies are following a middle path that combines the best of both worlds. Jeff has coined a term for this hybrid approach: cloudbursting.  

Jeff defines cloudbursting as “an application hosting model which combines existing corporate infrastructure with new, cloud-based infrastructure to create a powerful, highly scalable application hosting environment.” Here’s his description of how it works: 

A pattern is starting to emerge. The conservative side advocates keeping core business processes inside of the firewall. The enthusiasts want to run on the cloud. They argue back and forth for a while, and eventually settle on a really nice hybrid solution. In a nutshell, they plan to run the steady state business processing on existing systems, and then use the cloud for periodic or overflow processing.

This hybrid approach may not slow the interest in private clouds that employ cloud computing concepts to run apps more efficiently in corporate data centers, but it will raise awareness of different ways to manage the risks and rewards presented by third-party clouds.

Amazon Adds Elastic Block Storage (EBS)

Amazon Web Services has launched Elastic Block Services (EBS), adding persistent storage capabilities to its cloud computing service, expanding the utility and potential uses of Amazon’s platform.

Prior to Amazon EBS, storage within an Amazon EC2 instance was tied to the instance itself so that when the instance was terminated, the data within the instance was lost, Amazon said. With EBS, users can choose to allocate storage volumes that persist reliably and independently from Amazon EC2 instances. EBS also provides the ability to create point-in-time, consistent snapshots of volumes that are then stored to Amazon S3.

“Persistent block storage has been among the top requests of developers using Amazon EC2, and we’re excited to deliver Amazon Elastic Block Storage designed specifically for our cloud-based, elastic computing environment,” said Peter De Santis, General Manager of Amazon EC2.

“In short it’s a SAN (Storage Area Network) in the cloud,” explains the RightScale blog, which also offers an illustrated explanation of how EBS differs from previous Amazon services, and why it matters. “You can allocate a disk volume of 1GB to 1TB in size from what is now an endless SAN in the cloud and attach it to an instance of yours running in EC2.”

S3 Downtime: More Missing Bites

Amazon has issued a detailed timeline and explanation of last Sunday’s lengthy outage for its S3 utility storage service. Here’s the root cause:

We’ve now determined that message corruption was the cause of the server-to-server communication problems. More specifically, we found that there were a handful of messages on Sunday morning that had a single bit corrupted such that the message was still intelligible, but the system state information was incorrect.

Single-bit corruption was also a key issue in an S3 outage on June 20, in which a single load balancer was cited as the culprit in the file corruption, which affected customers using MD5 checksums to verify data integrity. After that incident, Amazon said it would “improve our logging of requests with MD5s, so that we can look for anomalies in their 400 error rates. Doing this will allow us to provide more proactive notification on potential transmission issues in the future.”

Amazon S3 Issues: Load Balancers and MD5

Amazon’s S3 storage system had some issues last week with data corruption on files using MD5 to perform integrity checks. After some investigation, Amazon confirmed the problems and identified the cause:

We’ve isolated this issue to a single load balancer that was brought into service at 10:55pm PDT on Friday, 6/20. It was taken out of service at 11am PDT Sunday, 6/22. While it was in service it handled a small fraction of Amazon S3’s total requests in the US. Intermittently, under load, it was corrupting single bytes in the byte stream. … Based on our investigation with both internal and external customers, the small amount of traffic received by this particular load balancer, and the intermittent nature of the above issue on this one load balancer, this appears to have impacted a very small portion of PUTs during this time frame.

There are several follow-ups of note: Alistair Croll at GigaOm takes a look at the role of load balancers in cloud platforms, while Craig Balding of Cloud Security takes a look at the MD5 issues.