[FoRK] FYI Netflix is down

Aaron Burt aaron at bavariati.org
Sun Jul 1 00:39:02 UTC 2012

On Sat, Jun 30, 2012 at 03:15:07AM -0400, Andrew D Kirch wrote:
> On 6/30/2012 3:11 AM, Tyler Haske wrote:
> >How to run a datacenter 101. Have more then one location, preferably
> >far apart. It being Amazon I would expect more. :/

Amazon has many datacenters and tries to make it easy to diversify.

> Based on?  Clouds are nothing more than outsourced responsibility.
> My business has stopped while my IT department explains to me that
> it's not their fault because Amazon's down <snip>

It *is* their fault.  You can blame faulty manufacturing for having a HDD
die, but it's IT's fault if it takes out the only copy of your database.

AWS 101:  Amazon has clearly-marked "Availablity Zones" for a reason.

Oh, and business 101: have an exit strategy for every vendor.

This outage is mighty interesting.  It's surprising how many big operations
had poor availability strategies.  Also, I've been working on an exit
strategy for one of my VM/colo providers, and AWS + colo in NoVa is one of
my options.

> The cloud may be a technological wonder, but as far as
> business practices go, it's a DISASTER.

I wouldn't say so.  Like any disruptive service, you're getting an
acceptably lower-quality product for significantly less money.  And like
most disruptors, it operates by different rules than the old stuff.

FoRK mailing list

More information about the NANOG mailing list