[Fiware-cloud] 16-Aug-2011 weekly conf call -- minutes & next steps

[Alex Glikson]

Dear all,

Please, see below a summary of our last weekly conf call. Your comments 
are welcome.

For our next calls, I would suggest the following agenda (aiming at 
finalizing the architecture, assets and gaps for each GE):
23/8 -- Object Storage GE and PaaS Management GE (Intel and TID are lead 
contributors, correspondingly)
30/8 -- Service Management GE and Portal (TID and UPM are lead 
contributors)
6/9 -- Cloud Edge GE (TRDF is the leading contributor)

For each of the topics, please, make sure that representatives from lead 
contributors attend and are well-prepared to discuss architecture, assets 
and gaps.

Regards,
Alex

Attendees: IBM, TID, INTEL
Agenda: architecture and assets for IaaS DataCenter Resource Management GE 
and Monitoring GE

IaaS Resource Management GE
Assets:
OpenStack Nova is currently the leading candidate to be the basis for IaaS 
resource management functionality. It is well-architected, and has a 
strong growing community contributing to it.
OpenStack Glance can be considered as a basis for image library 
functionality. However, the relationship to the asset proposed for service 
catalog managed by the Service Management GE needs to be clarified [owner: 
TID]. For example, does Glance need to keep OVFs, or just the individual 
VM images (while OVFs will be kept in Claudia)?
Open vSwitch is the leading candidate to provide virtual networking 
functions, such as support for VLANs (an integration with OpenStack 
exists, but might require effort to provide what we need)
PIVOT (from IBM) will be used for intelligent management (incl. resource 
scheduling, policies and admission control) of 'clusters' comprising the 
cloud infrastructure, potentially in a hierarchical manner (for 
scalability)
xCAT (provided by IBM) will be used for bare-metal provisioning and 
hardware management of the physical machines comprising the cloud 
infrastructure.
HSLT (from IBM) will provide resiliency features for the management 
fabric. 
Several tools can be considered to collect the monitoring metrics of the 
various resources, such as collectd, Ganglia, Nagios and xCAT. This would 
need to integrate with the framework provided by the Monitoring GE 
(described below).
OCCI will be considered as the standardized interface that will be 
surfaced by this GE
Other related assets to consider: Crowbar, Opscode (there are probably 
more)

Architecture:
The basis for the architecture will be the one of OpenStack Nova. A good 
summary can be found at 
http://ken.pepple.info/openstack/2011/04/22/openstack-nova-architecture/. 
TBD: need to map all the other assets on the architecture [owner: IBM]

Gaps:
Integration
QoS
admin kit
Other?

Monitoring GE:
Note: the discussions regarding this GE are rather preliminary at this 
stage.
An asset being considered for the monitoring GE is the Rocksteady 
framework (http://code.google.com/p/rocksteady/)
An architecture including also few additional components (RabitMQ, Esper, 
Graphite) to gather and process metrics is outlined at 
http://code.google.com/p/rocksteady/ 

Gaps:
TBD, requires further discussion
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.fiware.org/private/old-fiware-cloud/attachments/20110817/24ef0c97/attachment.html>