Disqus - Latest Comments for Jesse Robbins

Re: The cable cuts: Get out the foil hats

Jesse Robbins — Wed, 06 Feb 2008 21:48:59 -0000

Mathew,

Thank you for understanding the nuance in this situation and for helping spread the word. It's amazing how much steam this built up in such a short amount of time. After I made my first post on Saturday, I followed it up with a conspiracy post because I find these sorts of things are both interesting and possible. However I ended up editing it once the the reverb in the echo-chamber reached critical mass. It is particularly irritating that the digg and slashdot posts weren't meaningfully corrected once it was clear that the "iran is off the internet" headline simply wasn't true. If there is a conspiracy it's for website hits.

Re: Latest on data center 'incident'

Jesse Robbins — Sun, 05 Jul 2009 13:29:00 -0000

Using the term "Incident" to describe an event like this is completely correct.

Data Center failures happen, even in extremely well managed facilities, which is why experienced web site operators regard them as just another single point of failure.

The winning approach is not to try to eliminate these failures, but to accept that Failure Happens and then to engineer resilient systems & organizations to support them.

-Jesse Robbins

Re: Latest on data center 'incident'

Jesse Robbins — Sun, 05 Jul 2009 13:52:00 -0000

Additionally, I've written about a number of similar incidents over the past few years. You may find the following links helpful:

* http://radar.oreilly.com/archives/2007/11/failu...

* http://radar.oreilly.com/2009/04/att-fiber-cuts...

* http://radar.oreilly.com/2008/03/amazon-improve...

-Jesse Robbins

Re: How the Seattle data center fire caught companies unprepared

Jesse Robbins — Mon, 06 Jul 2009 09:41:00 -0000

@rick -

Absolutely, in many cases it is entirely acceptable to have a traditional DR strategy that relies on a single datacenter or provider. It is a business decision plain & simple, and a very important one.

-Jesse Robbins

Re: How the Seattle data center fire caught companies unprepared

Jesse Robbins — Mon, 06 Jul 2009 10:33:00 -0000

@rick

Thanks to virtualization & "Infrastructure as a Service" platforms like Amazon's EC2, it has gotten a lot easier to expand beyond one datacenter.

The primary challenges are:
1) Effective cross-datacenter data replication.
2) Building applications that handle higher-latency while still providing high-performance
3) Managing dynamic infrastructure effectively, regardless of if its in your own datacenter or in the cloud or somewhere in-between.

-Jesse Robbins
CEO, Opscode