Decomposed VNF monitoring framework

Registered by caoyue

Tacker now only support vnfm and nfvo is just started in progress ,for the
vnfm monitoring driver is ping and mgmt driver is ssh and infra driver is
via heat to call nova create vnf vm.
 we should detach an independent module from vnfm to complete the vnf
FCAPS(fault,configuration,accounting,performance,security) management.

As a first step we need to detach tacker monitoring framework to run as a
"tacker-monitor" worker process, support running multiple of them, possibly
in different servers and communicate with the main "tacker-server" process
using RPC. This will help towards HA and scalability of tacker VNF monitoring
framework.

From xiayu, "separate-monitor-from-tackerserver" BP:
tacker-server's monitor function has many problems. monitor can't be run in multi process, because vnf's add/ delete to monitor must be done in the same process. separate tacker-monitor from tacker-server is a solution to this problem., and can have better performance.

original title: design an em module from vnfm to complete the vnf fcaps management

Blueprint information

Status:
Complete
Approver:
Sridhar Ramaswamy
Priority:
Essential
Drafter:
yong sheng gong
Direction:
Approved
Assignee:
yong sheng gong
Definition:
Obsolete
Series goal:
None
Implementation:
Unknown
Milestone target:
None
Completed by
yong sheng gong

Related branches

Sprints

Whiteboard

The "ask" is too broad. Please help to narrow down the scope (sridhar_ram, 09/06/2016)

Updated BP with a specific reduced scope to decompose tacker monitoring framework out
of the main tacker server process
    - Sridhar Ramaswamy, 14-Dec-2016

Merged with xiayu's "separate-monitor-from-tackerserver" BP [*] and that is superceded into this BP.
    [*] https://blueprints.launchpad.net/tacker/+spec/separate-monitor-from-tackerserver
    - Sridhar Ramaswamy, 14-Dec-2016

@xiayu - Thanks for writing this one up. This is very useful / badly needed scale / performance enhancement for Tacker. Please join a tacker weekly irc meeting to discuss this further and help implementing this.
@xiayu - do you have plan to write a spec for this? thanks [yongsheng gong]

Moving to Pike - sridhar_ram, Feb 3, 2017

I have drafted a spec for it via GOOGLE DOC+ https://docs.google.com/document/d/1TQUyEek_M1jbYK_nqiUuPVLurx4LhRnpv7ZUCFHGM5M/edit?usp=sharing (gongysh)

Gerrit topic: https://review.openstack.org/#q,topic:bp/decomposed-monitor-framework,n,z

Addressed by: https://review.openstack.org/439300
    introduce a scalable task scheduling framework

Addressed by: https://review.openstack.org/443719
    Add basic task framework

(?)

Work Items

Dependency tree

* Blueprints in grey have been implemented.

This blueprint contains Public information 
Everyone can see this information.