July 21st, 2008
S3 outage: time to double up
Probably the best presentation at London Cloud Camp last Wednesday was the one the organizers saved until last: Alan Williamson spoke about Mediafed’s experiences as a company that relies on cloud providers. Mediafed specializes in providing RSS traffic analytics to European media companies, with a blue-chip client roster that includes BBC Worldwide, LeMonde, The Guardian, IDG, Axel Springer and others. Williamson’s advice will be heeded by many wondering what to do after Amazon S3’s 6-hour long outage yesterday:
“We’ve come to realize we cannot rely on putting all our eggs in one basket,” he said, explaining that Mediafed uses two cloud providers side-by-side: Amazon and UK-based Flexiscale. “We run both at the same time.” Some customers are on one and some on the other, but all are backed up to the other cloud so that if one fails the service can switch across to the other — which presumably means its customers can still this morning access all their stats from yesterday’s RSS traffic, even during the hours that S3 was down.
Anyone concerned about what Om Malik is calling the fragility of cloud services after yesterday’s outage needs to consider putting a similar set-up into place. Frankly, I think Malik has it completely wrong — it’s not the cloud that’s fragile, it’s computers, and anyone who expects perfect uptime when relying on a single point of failure has their head, not just their infrastructure, in the clouds. Either you stay cool, like SmugMug, and accept occasional glitches as part of the value proposition you pay for; or you do what Mediafed has done and put some redundancy and a failover plan in place.
As Williamson said in his presentation, there are plenty more risks to worry about besides systems outages: “My heart is fearful of the credit card stopping at Amazon. It scares the bejesus out of me that’s going to happen at Amazon.” For all its benefits, the cloud is still no silver bullet. Working with the cloud means getting savvy about a whole new set of issues, such as becoming expert in building server infrastructure that monitors your cloud resources, or solving hairy back-up and archiving challenges (Mediafed recently calculated that it is now storing so much data in the cloud that it would take three weeks to download a back-up of its S3 data). Williamson concluded: “We appreciate that cloud computing has moved us on but [now] we’ve got a whole set of new problems.”
See also these posts about previous outages at Amazon Web Services:
Amazon Web Services gets serious about enterprise
Time for a Bezos trustworthy cloud initiative?
Phil Wainewright is a commentator and strategist on emerging software industry trends. See his full profile and disclosure of his industry affiliations.
![[image]](http://mowser.com/img?url=http%3A%2F%2Fi.zdnet.com%2Fimages%2F200803%2FtreeItem.gif)
![[image]](http://mowser.com/img?url=http%3A%2F%2Fi.zdnet.com%2Fimages%2F200803%2FtreeLastItem.gif)
![[image]](http://mowser.com/img?url=http%3A%2F%2Fi.i.com.com%2Fcnwk.1d%2FAds%2F1671%2F12%2FIntel_PC08_eco.jpg)
![[image]](http://mowser.com/img?url=http%3A%2F%2Fi.i.com.com%2Fcnwk.1d%2FAds%2Fcommon%2Fcleargif.gif)
![[image]](http://mowser.com/img?url=http%3A%2F%2Fi.i.com.com%2Fcnwk.1d%2FAds%2F2442%2F12%2Fsuper-techies_120x90.jpg)
Checking your mobile data usage online is not easy [Geekzone: IT, mobility, wireless and handheld news]. Posted: 22 Jul 2008 03:24 AM CDT. Isn't that incredible that Telecom New Zealand can charge my mobile data usage every month but it ...
Trackback by apprize — July 22, 2008 @ 12:19 pm
The Wisdom of Cloudsof techniques for achieving this, such as redundancy and caching, but I will dive into those in more depth in a later post. (A great source for these concepts is http://highscalability.com.) Some of the true pioneers in the cloud realized this early. Phil Wainwright notes that Alan Williamson of Mediafed made what appears to be a prescient decision to split their processing load between two cloud providers, Amazon EC2/S3 and FlexiScale. Even Amazon themselves use caching to mitigate S3 outages on their retail sites
Trackback by Anonymous — August 16, 2008 @ 3:05 am
ElephantDriveof an outage or problem by either we can continue to provide uninterrupted service. This was put to the test again this weekend when the S3 service experienced a major outage. Details of the downtime were covered by a variety of sources, some of which detailed responses from end users. The good news is that our system worked flawlessly - our storage allocation engine recognizes the problem and routed requests to available nodes instead of the non-responsive S3 service points. We’
Trackback by Anonymous — August 16, 2008 @ 3:05 am
The CIO Weblog - Main page - Business and Technology Coverage of the Software Market - cio, erp, crm, enterprise, software, SOA, SaaSwas perhaps more overblown than I would have preferred it to be as I survey it in the cold, clear light of morning. A surprising number of pundits are dragging out the "don't put your eggs all in one basket" admonishment, including Phil Wainewright and Dave Winer. This is a bit disappointing as it seems to me at least to almost entirely miss the point of cloud computing; although at this point it may be good advice for anyone relying on Amazon in particular... but perhaps that just points to
Trackback by Anonymous — August 16, 2008 @ 3:05 am