Summary of comments following the USAG 20090402 meeting notes' publishing
http://indico.cern.ch/materialDisplay.py?materialId=minutes&confId=55267
=========================================================================

These comments are taken into consideration in the new OLA for TPMs 
http://indico.cern.ch/materialDisplay.py?materialId=0&confId=57766 draft.

* The TPM decentralisation advantages are:
	. more scalable 
	. closer to the users

* Neverteless, there will, probably, never exist as many TPM teams as EGI 
NGIs, so the decentralisation model will be full of exceptions.

* EGI.org must supervise and coordinate the NGI TPM teams and provide a 
central TPM role in addition.

* TPMs are not just ticket dispatchers but real Grid generalists. NGIs 
must build this knowledge or get others to provide this service for them.

* It is important to decentralise without fragmenting! 

* Dedicated, infrequent shifts will always be necessary.

* The most serious support problems are due to the, often, poor middleware 
quality, a situation likely to get worse in EGI as there is no channel to 
influence that. Glite figures by S. Burke: 997 patches in total, of which 
132 were rejected, 181 were obsoleted by a replacement and 435 are in
production).

* All the effort invested so far in monitoring and support are proof of 
failure for the Grid.

* Scheduling a EDG/EGEE I,II,III catharsis meeting would be useful.

* As the COD work is also being decentralised use common tools and 
procedures where possible.

------------------------------------------------------------------------

Comments mostly by Steven Burke, Ron, Jeff, Vera, Dimitris, Helene, Fred, 
John Gordon (in no particular order) assembled by Maria on 2009-04-28