I know the OSS product is meant to offer an AWS-compatible API, but
was wondering whether that meant typica could be pointed at another
cluster management endpoint and used to bring up machine images there?
1. It's hard to blame typica for fundamental design flaws in W3C XML
Schema Language. I think JAXB may be playing up too, being over-fussy
about xmlns values. This is a problem most WS-* stacks end up
encountering.
2. Looking at Waheed's notes, one of the problems is the AWS
authentication protocol using request time, and assuming the client's
clock and TZ settings are always valid
http://waheediqbal.wordpress.com/hack-typica-to-work-with-eucalyptus-14/
I've had problems there with VMWare images talking to AWS classic.
Ideally, the EC2 api should include a method to give the current date
time/TZ at the far end, information that the client can use in future
calls, so it doesn't have to get its own clock in order.
I just found out that project I'll be involved with this summer will
involve some interoperability between EC2 and a Eucalyptus (local)
cloud. So, the good news is that I'll get some up close and personal
time with this issue!