Cloud computer not immune from the clouds?

[DJofSD]

http://www.forbes.com/sites/anthonykosner/2012/06/30/amazon-cloud-goes-down-friday-night-taking-netflix-instagram-and-pinterest-with-it/

and related

http://www.forbes.com/sites/eliseackerman/2012/06/17/the-real-story-behind-the-aws-outage-and-what-it-means-for-investors/

Clearly, there are some IT managers which will have to answer some hard questions very soon.

If I was high level officer in any of these companies, I'd be chewing some ass right now and expecting a head or two on a platter. In this day and age, failure to provide seemless availability is unforgiveable.

[HUSKER55]

i am no guru by any standard. However, since it was mother nature that caused the problems and we are only talking about one hour, would you please explain the problem again? I am sorry, I don't get it.

[DJofSD]

Any kind of down time is a loss of revenues. The other issue is cloud computing is by its definition distributed with the primary benefit being availability.

Why go to cloud and endure the additional complexity and expense if you do not derive the promised benefit?

[wilderness]

Amazon-EC2, despite their size, allows some very "shady" practices for hosting clients (i. e. harvesters).

EC2 is well aware of these on goings and has tailored a bread-and-butter profit from these types of customers.

Of course EC2 offers a variety of services.

The outage issue is a two-fold blame. (not including Mother Nature).
1) As DJ explained that EC2 didn't have instantaneous re-routing for such a large customer.
2) Shame on NetFlix for relying on a hosting company for service assurance and re-routing.

[DJofSD]

The outage issue is a two-fold blame. (not including Mother Nature).
1) As DJ explained that EC2 didn't have instantaneous re-routing for such a large customer.
2) Shame on NetFlix for relying on a hosting company for service assurance and re-routing.

Right and I agree.

A bit more background for 55-man: a large unstated assumption that goes to the heart of a modern data center and its operation is the elimination of all single points of failure (SPoF). A SPoF is exactly what the phrase implies: this piece breaks or that function is not available you experience a failure. This can be hardware or software, power, networking or any other service which is critical to delivery of IT services.

And, while seemingly over the top, even the data center itself can be considered a SPoF. This leads to a concept I learned to described as geographically diverse. And, the diversity applies to the networking connections the data center is dependent upon.

So, in my estimation, the failure experienced Friday night by those companies and their clients was foreseeable and preventable, especially if the company is touting cloud services.

[GameTheory]

2) Shame on NetFlix for relying on a hosting company for service assurance and re-routing.Netflix is currently building their own network/data centers, but up to now has relied on third-party providers for their streaming service.

[wilderness]

Netflix is currently building their own network/data centers, but up to now has relied on third-party providers for their streaming service.

They could still rely successfully on thrid-party servers IF they controlled and configured their DNS to secondary (BU) servers, which is what DJ and I were referring to.

[Dave Schwartz]

Here's to redundancy!

[Tom]

Here's to redundancy!

[DJofSD]

Here's to redundancy!
You can say that again!

[GameTheory]

Here's to redundancy!...with apologies to our British friends. (In which case it would mean, "Here's to being laid off!")

[HUSKER55]

THANKS DJ FOR EXPLAINING THAT FOR ME! :)

[DJofSD]

I'm happy to help. I'm not the biggest techno-geek out there but it has been a part of my life for a long time. Must be in my blood.