HI Some thoughts:
> Comparing the designs it seems that one distinguishing factor is that > OpenNebula has a central cloud controller which is a potential single point > of failure and potentially a bottleneck when scaling the size of the cloud > (don't know if this is an issue in practice). Nova design seems to be more > loosely coupled (but I might be wrong on this, as information on the web > about Nova at this stage does not seem always easy to understand or be > consistent with other information you find). Single point of failure: As far as I know the Nova/Eucalyptus Cloud-Cluster controller architecture does not provide any HA /failure tolerance feature. In this sense the Cloud controller is the single point of failure (I may be wrong, though). There are some HA work undergoing for the next release of OpenNebula, as well as some configurations that can be applied now to improve the failure tolerance of OpenNebula. Scalability: Note that the systems that will impact scalability (i.e. monitoring, image distribution mainly) does not need to follow a single tier architecture. In fact large scale OpenNebula deployments uses specialized monitoring systems (e.g Ganglia which is multi-tier) or high-formance distributed FS. Additionally, we are working in some OpenNebula-2-OpenNEbula drivers to further federate cloud infrastructures. Although the goal of this developments are not scalability but datacenter federation. Cheers Ruben -- Dr. Ruben Santiago Montero Associate Professor (Profesor Titular), Complutense University of Madrid URL: http://dsa-research.org/doku.php?id=people:ruben Weblog: http://blog.dsa-research.org/?author=7 _______________________________________________ Users mailing list Users@lists.opennebula.org http://lists.opennebula.org/listinfo.cgi/users-opennebula.org