Open Stack

On 01/15/2014 02:40 PM, Tim Bell wrote:
> 
> It's great to be having the discussion since we need to find what can be simplified and what is currently needed.
> 
> At one time, we had discussed some form of opt-in stats monitoring of production deployments (https://blueprints.launchpad.net/oslo/+spec/opt-in-stats-tracking). While there is no implementation, it would be a place where a set of counters could help out to assess the potential impact. There are lots of APIs (at CERN, we use the ones listed in http://information-technology.web.cern.ch/book/cern-cloud-infrastructure-user-guide/advanced-topics/programming-apis)
> 
> The challenge is to know which external facing APIs 
> 
> - are using XML and know they're using XML
> - think they are not using XML but actually one of their library/ecosystem products use it (it is quite difficult to know without diving into the logs)
> - weren't sure on the details of the question in the survey
> 
> For many sites, they use a library/ecosystem product and do not know what is underneath. Given a transition period, there is a good chance to avoid downtime.

There would definitely be a good transition period.  We're only talking
about doing this in a new major revision of the compute API (v3), which
will contain a number of non-backwards compatible changes, anyway.
There will be a transition period from v2 to v3.

-- 
Russell Bryant