gray area: services developed at slac might rely on info stored by NCSA layer (eg is the file on tape or disk)
VMs
IPAC needs few VMs @NCSA in the near future
for now: provision manually
Xiuqin will write down needs
API for cutout service
Brian's experience: python is easier if it is first experience. If need performance java is better, it has great parsing libraries, serializing/deserializing different formats in java is convenient. Django is fine for small applications. Framework "flask" is good too.
Ruby? Great to get started. Issues with performance, not best in long run.
fits libraries we rely on: afw, easily accessible from python
looks like python is a good choice for us
but check with Serge Monkewitz, he used it for IRSA, wasn't happy with some parts
be consistent, use the same API for all services
DDT for qserv-served tables?
proper solution coming in S15
for now, use the tables that are maintained for schema browser
User upload tables
come up with a plan during S15
Dataset/repo terminology
IPAC todo from past meetings, IPAC team will look into it
No meeting next week, it i a holiday at most labs.