After yesterday’s seven-plus-hour outage of both the US and UK servers, users of Amazon S3 are waking up with some difficult questions to answer.
While Amazon itself doesn't suffer 'outages' the S3 service has been down twice for significant periods of time in the past six months. Users have counted on Amazon's reputation, a 99.999 SLA, and engineered-in redundancy in using the service for 'in the cloud' storage and delivery of mission critical files. The relative ease of integration, the low cost, and the widespread adoption of S3 by startups has given Amazon an early lead in the storage community.
Now - after a day of broken pages, failed uploads, and cascading site failures, webmasters are determining if Amazon plans to build a service that only works for sites with low reliability requirements and startup budgets. Today, the broken sites include WordPress, Huffington Post, SmugMug, 37 Signals (Campfire, Backpack), CenterNetworks, NASDAQ, Activision, Business Objects, Hasbro and Magnify.net are just some of the many sites and companies using Amazon’s S3 Web Services.
The impact of the outage was so significant that Fred Wilson, traveling with his family in Europe twittered yesterday: "When I got on line yesterday I thought the Web was slow, until I realized that S3 was down." Oh, yeah, Twitter uses S3 too, so they were impacted. (Vator.tv was affected as well)
Amazon needs to answer the following questions quickly - and significantly. This can't be a press release. Jeff Bezos needs to publicly take responsibly for this outage, explain what happened, and clarify that Amazon expects to provide "Zero Defect" product to enterprise level customers. Anything less will signal the beginning of a defection of major customers from S3 to more expensive, but more reliable alternatives.
1. What happened.
The outage happened Sunday Morning at 9am PST, which looks more like a maintenance window than a time of high customer demand. Was this a self inflicted wound? We need to know.
2. Why won't it happen again?
It seems like there are some fundamental engineering issues with the service that make their promises of redundancy ineffective.
3. Why doesn't, and when will Amazon.com be reliant on S3 for its cloud computing /storage needs?
4. Is Amazon committed to powering web infrastructure? Simply put, is this a business or a hobby?
All of these questions come down to this: Will Jeff Bezo's eat his own dogfood? It's not a simple question. Someone needs to take ownership of the S3 business, clarify its commitment to customers, and layout a roadmap for the evolution of the product from start up haven to serious business computing platform. To do any less will signal to serious users that its time to look for other solutions and platforms. All of the hard work and innovation that Amazon has provided will quickly evaporate if they can't grow the service to meet the needs of their now deeply integrated customers.
This is a critical time for both S3 and the cloud computing movement. To call it anything less is to let the opportunity for leadership slip away.