Open Stack

Hi - 

After discussion with Jay Pipes about this bug
(https://bugs.launchpad.net/ceilometer/+bug/1211015), I'd like to split
out a new TraitType table in the storage layer, and remove the UniqueName
table. A TraitType is the name of the trait and the data type (i.e.,
string, int, floatŠ). I'd like to get some input on this change - maybe
it's over kill? Here is my rationale for adding this table:

1) The current query to return all trait names is slow, as stated in the
bug report.
3) All instances of event X will have the same traits, and all instances
of a trait have the same trait type name and data type. I think it is
cleaner to model this relationship in the db with an explicit TraitType
table.
2) The api needs a model for trait types in order to fulfill the
/v2/event_types/Foo</trait_type> query. This call will return the set of
trait names and data types, but no trait data.

Related patches:
sqlalchemy layer: https://review.openstack.org/#/c/42407/ (not sure the
migration is correct)

storage layer: https://review.openstack.org/#/c/41596/


Thanks!
-john
-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 5443 bytes
Desc: not available
URL: <http://lists.openstack.org/pipermail/openstack-dev/attachments/20130817/c352aae7/attachment.bin>