Sage Weil
2014-10-11 00:44:27 UTC
Here's my list of possible blueprints for hammer. And who might write
them up. This is just a brain dump of what I'm thinking about; please
add, remove, object, suggest, etc.!
Some of these blueprints already exist, but I suspect they are in need up
a refresh to make them reflect the current thinking.
I threw in several of the items we discussed in the performance calls. I
think most of the key areas of development are captured below. What's
missing is a lot of non-specific work around profiling and optimizing
existing code paths that is also a key effort for hammer.
What else is missing (or unnecessary) here?
-- core --
osd: recovery performance: EC recovery offload (Sam)
osd: qos: unified priority queue (Sam)
osd: performance: transaction api changes, keyfilestore backend (Sage)
osd: performance: rados io hints (?)
osd: data integrity: opportunistic at-rest checksums (#9059 Sage)
osd: data integrity: scrub and repair (David)
performance: buffer encode/decode performance (Matt)
performance: AsyncMessenger (Haomai)
performance: XioMessenger update (Matt)
mon: performance: improve threading/locking (Joao, Sage)
mon: scalability: prepopulate pg_temp (#7173 Sage)
mon: data integrity: per-crush type down_out_interval (#7347)
mon: data integrity: usable mon scrub (Joao)
auth: kerberos support (Sage)
tracing roadmap discussion
crush: multi-type trees (ssd hdd) (Sage)
-- librados --
librados: data integrity: end-to-end checksums (expose csums to read/write) (Sage)
librados: threading and async ops vs python (Dan)
-- cephfs --
data integrity: forward scrub (Greg)
quotas vs subtrees
auth (kerberos) minimum requirements
hadoop: status, progress, roadmap (Noah)
-- rgw --
rgw: api fidelity: object versioning update (Yehuda)
rgw: scaling, performance: bucket index scaling update (Guang)
rgw: multi-dc status, roadmap (Yehuda)
rgw: librgw?
rgw: api fidelity: object expiration
-- rbd --
rbd: performance: object map (Jason)
rbd: mirroring (Josh)
rbd: HA iscsi gateways (Mike)
rbd: performance optimization discussion
-- calamari --
rpm/deb packaging update for calamari and deps (Boris, Tim, Dan?)
kerberos auth support (Dan)
calamari api gaps (e.g., rbd) (Gregory)
-- calamari-clients / romana --
localization update?
calamari vs romana roadmap discussion
-- teuthology --
teuthology: general roadmap discussion
paddles: queueing ?
automated performance tests
-- ceph-deploy --
?
-- build/ci --
build infrastructure roadmap
--
To unsubscribe from this list: send the line "unsubscribe ceph-devel" in
the body of a message to ***@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
them up. This is just a brain dump of what I'm thinking about; please
add, remove, object, suggest, etc.!
Some of these blueprints already exist, but I suspect they are in need up
a refresh to make them reflect the current thinking.
I threw in several of the items we discussed in the performance calls. I
think most of the key areas of development are captured below. What's
missing is a lot of non-specific work around profiling and optimizing
existing code paths that is also a key effort for hammer.
What else is missing (or unnecessary) here?
-- core --
osd: recovery performance: EC recovery offload (Sam)
osd: qos: unified priority queue (Sam)
osd: performance: transaction api changes, keyfilestore backend (Sage)
osd: performance: rados io hints (?)
osd: data integrity: opportunistic at-rest checksums (#9059 Sage)
osd: data integrity: scrub and repair (David)
performance: buffer encode/decode performance (Matt)
performance: AsyncMessenger (Haomai)
performance: XioMessenger update (Matt)
mon: performance: improve threading/locking (Joao, Sage)
mon: scalability: prepopulate pg_temp (#7173 Sage)
mon: data integrity: per-crush type down_out_interval (#7347)
mon: data integrity: usable mon scrub (Joao)
auth: kerberos support (Sage)
tracing roadmap discussion
crush: multi-type trees (ssd hdd) (Sage)
-- librados --
librados: data integrity: end-to-end checksums (expose csums to read/write) (Sage)
librados: threading and async ops vs python (Dan)
-- cephfs --
data integrity: forward scrub (Greg)
quotas vs subtrees
auth (kerberos) minimum requirements
hadoop: status, progress, roadmap (Noah)
-- rgw --
rgw: api fidelity: object versioning update (Yehuda)
rgw: scaling, performance: bucket index scaling update (Guang)
rgw: multi-dc status, roadmap (Yehuda)
rgw: librgw?
rgw: api fidelity: object expiration
-- rbd --
rbd: performance: object map (Jason)
rbd: mirroring (Josh)
rbd: HA iscsi gateways (Mike)
rbd: performance optimization discussion
-- calamari --
rpm/deb packaging update for calamari and deps (Boris, Tim, Dan?)
kerberos auth support (Dan)
calamari api gaps (e.g., rbd) (Gregory)
-- calamari-clients / romana --
localization update?
calamari vs romana roadmap discussion
-- teuthology --
teuthology: general roadmap discussion
paddles: queueing ?
automated performance tests
-- ceph-deploy --
?
-- build/ci --
build infrastructure roadmap
--
To unsubscribe from this list: send the line "unsubscribe ceph-devel" in
the body of a message to ***@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html