config: improve performance by not reading backref/children from db if not needed

[Hampapur Ajay]

in cassandra_<type>_read() implementation of autogen code, all the columns of an object are read from DB and filtered by field_names by iteration.

In scale situations this is non-performant as project, ipam, sg etc. have many backrefs and most common case is to not need backref/children of these objects.

So fix autogen code to check required fields against the set of backref + children and read-in with column_start if not only props/ref are needed.

[OpenContrail Admin (ci-admin-f)]

Reviewed: https://review.opencontrail.org/7438
Committed: http://github.org/Juniper/contrail-generateDS/commit/d1c215fd9cab5a74edd698f41ce1bc361bc2eb7a
Submitter: Zuul
Branch: R2.0

commit d1c215fd9cab5a74edd698f41ce1bc361bc2eb7a
Author: Hampapur Ajay <email address hidden>
Date: Fri Feb 13 10:10:33 2015 -0800

config: do not read backref/children fields from cassandra if not needed

from the bug descr:
in cassandra_<type>_read() implementation of autogen code, all the columns
of an object are read from DB and filtered by field_names by iteration.

In scale situations this is non-performant as project, ipam, sg etc. have
many backrefs and most common case is to not need backref/children of these objects.

So fix autogen code to check required fields against the set of backref + children
and read-in with column_start if not only props/ref are needed.

Change-Id: Ie96744e829f67409c6cb76deecb4e7117ac61d5d
Closes-Bug: #1421752
(cherry picked from commit acc3904c3c0b56e9871d885c24ad5404814e4e8a)

[OpenContrail Admin (ci-admin-f)]

Reviewed: https://review.opencontrail.org/7433
Committed: http://github.org/Juniper/contrail-generateDS/commit/38d7f19de3fb5ecde332f37a0d4fd6ea0d971e31
Submitter: Zuul
Branch: R2.1

commit 38d7f19de3fb5ecde332f37a0d4fd6ea0d971e31
Author: Hampapur Ajay <email address hidden>
Date: Fri Feb 13 10:10:33 2015 -0800

config: do not read backref/children fields from cassandra if not needed

from the bug descr:
in cassandra_<type>_read() implementation of autogen code, all the columns
of an object are read from DB and filtered by field_names by iteration.

In scale situations this is non-performant as project, ipam, sg etc. have
many backrefs and most common case is to not need backref/children of these objects.

So fix autogen code to check required fields against the set of backref + children
and read-in with column_start if not only props/ref are needed.

Change-Id: Ie96744e829f67409c6cb76deecb4e7117ac61d5d
Closes-Bug: #1421752
(cherry picked from commit 5c674d0bd5b3575bf4dd3a42b5c1fa469cb226bb)

[OpenContrail Admin (ci-admin-f)]

Reviewed: https://review.opencontrail.org/7799
Committed: http://github.org/Juniper/contrail-generateDS/commit/13ab69f5fc17de340cddbcaa8f3ac1734cdd2b82
Submitter: Zuul
Branch: R1.10

commit 13ab69f5fc17de340cddbcaa8f3ac1734cdd2b82
Author: Hampapur Ajay <email address hidden>
Date: Fri Feb 13 10:10:33 2015 -0800

config: do not read backref/children fields from cassandra if not needed

from the bug descr:
in cassandra_<type>_read() implementation of autogen code, all the columns
of an object are read from DB and filtered by field_names by iteration.

In scale situations this is non-performant as project, ipam, sg etc. have
many backrefs and most common case is to not need backref/children of these objects.

So fix autogen code to check required fields against the set of backref + children
and read-in with column_start if not only props/ref are needed.

Change-Id: Ie96744e829f67409c6cb76deecb4e7117ac61d5d
Closes-Bug: #1421752
(cherry picked from commit acc3904c3c0b56e9871d885c24ad5404814e4e8a)
(cherry picked from commit d1c215fd9cab5a74edd698f41ce1bc361bc2eb7a)

[OpenContrail Admin (ci-admin-f)]

Reviewed: https://review.opencontrail.org/7421
Committed: http://github.org/Juniper/contrail-generateDS/commit/48550eff29ec91d3645ee678276dc712c340e7a3
Submitter: Zuul
Branch: master

commit 48550eff29ec91d3645ee678276dc712c340e7a3
Author: Hampapur Ajay <email address hidden>
Date: Fri Feb 13 10:10:33 2015 -0800

config: do not read backref/children fields from cassandra if not needed

from the bug descr:
in cassandra_<type>_read() implementation of autogen code, all the columns
of an object are read from DB and filtered by field_names by iteration.

In scale situations this is non-performant as project, ipam, sg etc. have
many backrefs and most common case is to not need backref/children of these objects.

So fix autogen code to check required fields against the set of backref + children
and read-in with column_start if not only props/ref are needed.

Change-Id: Ie96744e829f67409c6cb76deecb4e7117ac61d5d
Closes-Bug: #1421752