Livefyre Profile

Activity Stream

Well, how bad was Azure's storage performance that it required an update that appears to have been mistakenly applied to multiple regions at once? One issue is the actual fix, which appears not to have been fully tested. Another issue is the procedure for deploying the fix, which appears to have been done improperly or not according to procedure. The result is another "black eye" for Microsoft Azure. AWS has been down this road too. When operating a service at web-scale, there is a huge downside if a mistake or error can bring down the service. In theory, the service should be architected to contain a failure condition by having sufficient check-points or "safety valves" set to halt a widespread service outage. The post mortem after one of these outages usually results in revised procedures or improvements in "fail-safe" mechanisms. Running a service at web-scale is still challenging because it has not been done for a very long time. AWS has been at it for 8 years and Microsoft has been at it for about 5 years. That said, routine procedures, like pushing out updates, should never bring the service down. Given Mr. Zander's comments about this particular Azure outage, the architecture of the service should be analyzed and improved to prevent similar events.  Apologies do not fix anything.

5 hours, 4 minutes ago on Azure Outage

Reply

Makes sense to me.  I've advocated for years that unless you can agree on a definition of cloud computing your conversation can get confusing and unproductive.  The NIST definition of cloud computing is a good starting place since it is vendor neutral and NIST is a respected organization. Coming up with an easy to remember 5-3-2 makes it easy to recall the elements of the NIST definition. So, if you ever find yourself speaking with someone who either thinks you can define cloud computing anyway you want or there  is no definition of cloud computing, just repeat 5-3-2 and make them go away.

1 year, 7 months ago on The 5-3-2 Principle Of Cloud Computing: An Easier Approach

Reply

Interesting to see that Google has yet to create their own platform and get that ecosystem thing working for them. With competitors like Amazon, Oracle, Microsoft and Salesforce.com all using platforms to build their developer ecosystems, it was shocking to see the lack of appreciation for creating a platform at Google that any 3rd party developer can plug into. Well, I guess if search is your business, maybe the idea of creating a platform for everything to plug into might not have seemed obvious, but shouldn't we have expected more from a company that was built on questioning the status quo? Hey, aren't the smartest "guys" out there working at Google?<g>

3 years, 1 month ago on Google Engineer Accidently Shares His Internal Memo About Google + Platform

Reply