By: KenLC

KenLC — Fri, 26 Oct 2012 20:58:34 +0000

A couple of points about the Amazon outages:1. The June outage was technically caused by nature, but poor management “allowed” it to happen. If they had switched the diesels manually 30-45 minutes before the storm hit, they may not have had any outage. Any good data center manager knows to go to diesel BEFORE severe weather hits. Human error?NOTE: Microsoft had a major outage when they made a major network change on a Monday morning at 11 am. Enterprise data centers (and traditional outsourcing providers) do not make changes during primary business hours. Human error?2. The level of communications with users during an Amazon outage has been horrible. During the June outage, there were 3-4 hour gaps in status updates. This would NEVER be the case in an enterprise’s own data center. Amazon (and the other mega-cloud providers) need to address this gap, even if it is an additional charge for “Platinum customer service”, where clients can get guarenteed status updates every 15/30/60 minutes.3. More of a question: have the mega-cloud providers allowed their “shared” environment to simply become too large? Amazon had availability zones, but several outages have affected clients in multiple availability zones. Should they be building a series of isolated smaller shared cloud environments? I believe several outages have been exacerbated by replication storms, where the automated recovery overwhelmed their networks. I view most of these issues as a lack of maturity in the mega-cloud provider world. They need to actively look at how large enterprises and traditional outsourcing providers run their environments. One problem is that they are mixing consumer, SMB environments with larger enterprise, mission-critical environments and treating them with the same priority. The impacts are dramatically different and these providers need to understand the differences and differentiate their services.I have every faith that the Cloud Computing world will get there….eventually. Until progress is made, I believe larger enterprises will keep their mission critical environments in-house.

By: Christine Herbert

Christine Herbert — Wed, 24 Oct 2012 17:15:42 +0000

I think this may relate to the large solar flare that came our way a couple of days ago. See here: http://www.reuters.com/video/2012/07/13/large-solar-flare-signals-weekend-disrup?videoId=236482140

Comments on: An Outage is an Outage

By: KenLC

By: Christine Herbert