Currently VDSM gets polled for a lot of data from the engine. A lot of this data does not need to be transmitted. Currently we have the following calls * Every 3 seconds `list` * Every 15 seconds `getAllVmStats` The call to getAllVmStats can easily grow to several hundreds of kilobytes per call. However the majority of the data sent by VDSM to the engine is not required by the engine at all. However since there is currently no more effective way the engine is required to query the data all the time. Goals for this RFE: Reduce the VM data transfer size to a minimum and provide means to get only data which did really changed since the last request.
didn't make much progress yet...
dropping 3s interval by implementing events over JSON-RPC (see http://gerrit.ovirt.org/#/q/status:open+project:vdsm+branch:master+topic:events_vm_status)
moving over there *** This bug has been marked as a duplicate of bug 1225368 ***