A less than merry Christmas for Netflix
Summary: Yet another Amazon cloud failure; will anyone notice?
If you were one of those people whose Christmas plans involved a serious overdose on Christmas themed movies via your subscription to Netflix, I’m sure you were disappointed settling for whatever movies were available on cable or OTA, as Netflix experienced a back-end failure that shut the service down Christmas Eve through Christmas day.
Once again, the outage wasn’t specifically Netflix’s fault, other than in their choice of back-end service provider. Hosted on the Amazon cloud, the Netflix outage was a result of problems with the Amazon service. Amazon reports the problem was with their Amazon Web Services Elastic Load Balancer, in the US-East-Region1 datacenter.
Netflix is one of Amazon’s most highly visible customers for their cloud backend services, is likely the one that generates the highest amount of traffic, and which has a very sophisticated Content Delivery Network. In short, it’s a very high-profile customer and this is the second time that an Amazon failure has brought down Netflix service delivery in the last six months. Netflix has done what they can to ameliorate the effects of back-end problems, with their own CDN servers that are being deployed at hub ISPs through-out their service area, but the Amazon cloud failures have still shut down their service for extended periods.
Though its timing is unlikely to make it noticed by business customers, the scale of the failure, in bringing down the Netflix service, needs to be taken into consideration when businesses begin to transition more real-time 24/7 business applications to cloud backends. The question remains if these failures will be taken as a reflection on the cloud services industry as a whole or whether they will be laid at the feet of problematic design issues within the Amazon Web Services.
Kick off your day with ZDNet's daily email newsletter. It's the freshest tech news and opinion, served hot. Get it.
Talkback
it failed because Netflix is proprietary
Wrong
Well,
Re: Well
With HP-UX one year, I was on a call for over 20 hours to repair a development box after a patch of theirs made the system unusable. The only reason I even called them as I was told to by a "superior" as that why we paid so much for support. Every time they had a shift change we started over. When the "superior" came back into work the next morning I got permission to end the call and restore service my way (less than 15 minutes). Needless to say I ended up being their boss a few years later. Our dependency on software vendors in mission critical spaces dropped substantially.
Planning is everything. Customers don't care why a service is down. The service provider is responsible for their service in the eyes of the customer. A company pointing the finger at anyone else for their service being down is a sign of immaturity and customers see that.
Netflix's service being down is their fault. They take money from their customers to provide a service, how they provide that service is their responsibility. The fact that Netflix did not plan, or make sure that their provider (Amazon) had an adequate plan is Netflix's problem.
Re: those clowns couldn't fix a piece of toast.
Yeah right
Got some fact to back up your assertions
Have you bother to do a scrap of analysis to prove your position and the huge problem it would face. Not only the problem is beyond in the scope of open source, the people probably doesn't even have the tools or willingness to fix the problem for free. If I had the skills to fix Netflix problem I want to be paid for my work.
Then there huge security risk not only for theft but to attack other computers.
no, the real problem is
Sorry it can have viruses
Uhm...
Please
Solaris. That is a dead O/S at this point for many reasons. I remember a bug back in the Sun days for JAVA on Solaris that nailed us. The funny thing is that the bug had been known about for a few years and fixed on Windows, and if memory serves Linux as well. But Solaris...no. The bug that hit us with one of Sun's utilities that used JAVA to manage...get this...Solaris.
Oracle has only made the situation worse. Believe it or not, they have a class that they offer to their customers to better navigate Oracle's support. Ridiculous. If they have to offer a class to use their support system they might want to rethink how their support actually works. Aside from Oracle's database the fortune 150 company that I work for has been moving away from Oracle's products. Now we do have a plan to move away from their database if their costs get out of line, or the product is not longer being viewed helpful. We already have move away from dataguard for our RAC environments and replaced it with in-house code and disk cabinet based replication.
That's funny (not)
Not to mention that the problem was with *Amazon's* servers... you know, the company that uses the *Android* version of Linux for its tablets.
Android
Netflix isn't the problem. Amazon is.
Or break it as usual....
Amazon's streaming video worked fine...
My thought as well
Amazon new, Netflix old.
Was the "upgrade" issue just on 1 device?
Nor did it affect using their website, using the iOS app on an iPod, or using it on a Roku, etc.
Maybe Amazon just didn't want a competitor's service to run well on their product...
Then you're not very smart