Mohammad Shahbaz Memon
(Forschungszentrum Juelich GmbH (DE))
Mischa Salle (NIKHEF)
Oscar Koeroo (NIKHEF)
Björn Hagemeier (JUELICH)
Shahbaz Memon (JUELICH)
Laurence Field (CERN)
Zsombor Nagy (NIIF)
- main focus of this meeting was to narrow down this group's focus to
avoid catering broader usecases
which are unrealizable within EMI lifetime, and not to have double
effort with other projects
- What kind of use cases we should be focusing on?
- Laurence suggested to streamline our work in three parts
- introduction to the virtualization and cloud technologies –
background and state of the art.
- short and medium term goals (SMG) or items which we think are our
low hanging fruit
- long term goals not necessarily be implemented in the project
lifetime, but they could be used for the next generation of EMI in the
form of position/policy statements. These statements can be iterated
by the time we learn from the realization of short / medium goals
- Further notes on SMG
- they should define things in simplified user stories, such that
they can be taken as a requirement or a feature request to the EMI
- the possible source are our well-known customers and also the
technology folks developing in EMI
- one foreseeable obstacle is to decide which target environment
should be considered for our implementation
- Gap analysis is in the scope of EGI, rather than EMI.
- As in our group we already have a representation from the compute
and security areas, remaining participation is from the data area. The
group will contact area lead to help.
- StratusLab Collaboration Effort
- Björn recently created a unicore-6.4.1 over centos (single-site)
virtual appliance, that can easily be deployed and executed on the
stratus lab cloud. The appliance has been tested with simple job
- Zsombor will produce ARC's virtual appliance for StratusLab
environment. It will be optimal to have a basic appliance by EGI TF.
- As StratusLab has already deployed gLite based virtual
appliances, we may possibly leverage their experiences to gather any
requirements or short comings. Furthermore, we could also communicate
with WnodeS in this regard.
- How will Argus react to the virtualized middleware services?
Theoretically it should work in a seamless manner no matter if the
services are short lived; practically it has been not tested yet.