Personally, I can see there are several criteria for the replacement project(s) of Ceilometer/Aodh:

- Provide consistent API experience with other openstack services.
- Integrate with Keystone
- Have ALL the metrics of core services that Ceilometer is collecting nowadays.
- Alarm API
- Including all the feature-parity functionalities of current evaluators/notifiers Aodh already provides.
- Last but not least, easy to integrate with other OpenStack services

Only the projects that have all the answers "YES" is to be considered as an option.

Gnocchi is another story, it provides storage(by leveraging storage backends) function, aggregation function and user-facing API, there is no hard dependency between Gnocchi and Ceilometer, it could be replaced only if the alternative is what we expected.

---
Cheers,
Lingxian Kong
Catalyst Cloud


On Wed, Apr 10, 2019 at 8:04 PM Trinh Nguyen <dangtrinhnt@gmail.com> wrote:
Hi team,

Yesterday, we talk about Gnocchi and how it is embedded in many parts of Telemetry these days [1]. Since we many things in Ceilometer have been replaced by Gnocchi which is an external project, we need the project to be sustainable at some point. I'm not sure about that since the Gnocchi's cores change jobs recently (?). So, I continue here the discussion on whether or not to find alternatives for Gnocchi. People have suggested:
  • Prometheus: we then can get rid of AODH and use Prometheus's Alert Manager
  • InfluxDB (part of Monasca): and because many parts of Monasca can do what Ceilometer/AODH doing we can possibly replace Telemetry completely (?)
Correct me if I'm missing something.

All ideas are welcomed.

Yours,

[1] https://etherpad.openstack.org/p/telemetry-train
[2] http://eavesdrop.openstack.org/irclogs/%23openstack-telemetry/%23openstack-telemetry.2019-04-09.log.html#t2019-04-09T06:36:15

--
Trinh Nguyen