[openstack-dev] [Ceilometer] PTL Candidacy
eglynn at redhat.com
Tue Sep 23 09:10:07 UTC 2014
I'd like to continue serving as Telemetry PTL for a second cycle.
When I took on the role for Juno, I saw some challenges facing the
project that would take multi-cycle efforts to resolve, so I'd like to
have the opportunity to see that move closer to completion.
Over Juno, our focus as a project has necessarily been on addressing
the TC gap analysis. We've been successful in ensuring that the agreed
gap coverage tasks were completed. The team made great strides in
making the sql-alchemy driver a viable option for PoCs and small
deployments, getting meaningful Tempest & Grenade coverage in place,
and writing quality user- and operator-oriented documentation. This
has addressed a portion of our usability debt, but as always we need
to continue chipping away at that.
In parallel, an arms-length effort was kicked off to look at paying
down accumulated architectural debt in Ceilometer via a new approach
to more lightweight timeseries data storage via the Gnocchi project.
This was approached in such a way as to minimize the disruption to
the core project.
My vision for Kilo would be to shift our focus a bit more onto such
longer-terms strategic efforts. Clearly we need to complete the work
on Gnocchi and figure out the migration and co-existence issues.
In addition, we started a conversation with the Monasca folks at the
Juno summit on the commonality between the two projects. Over Kilo I
would like to broaden and deepen the collaboration that was first
mooted in Atlanta, by figuring out specific incremental steps around
converging some common functional areas such as alarming. We can also
learn from the experience of the Monasca project in getting the best
possible performance out of TSD storage in InfluxDB, or achieving very
high throughput messaging via Apache Kafka.
There are also cross-project debts facing our community that we need
to bring some of our focus to IME. In particular, I'm thinking here
about the move towards taking integration test coverage back out of
Tempest and into new project-specific functional test suites. Also the
oft-proposed, but never yet delivered-upon, notion of "contractizing"
cross-project interactions mediated by notifications.
Finally, it's worth noting that our entire community has a big
challenge ahead of it in terms of the proposed move towards a new
layering structure. If re-elected, I would see myself as an active
participant in that discussion, ensuring the interests of the project
are positively represented.
More information about the OpenStack-dev