DB: Optimize update methods

Many of our resources' update methods read the resource from the DB,
joining other tables in some cases, and then do the update using
SQLAlchemy ORM engine before returning updated ORM object.

There are some resources -volume_attachment, snapshot, backup, CG,
CGSnapshot, qos_specs, volume_type, and group_type- that don't care
about the updated ORM object, so we are actually doing unnecessary DB
operations.

This patch removes the unnecessary DB read operations when updating
those resources.

In the case of volume_type and group_type we are also retrieving a DB
entry instead of doing an exists check to check if a name chage is
valid.  This patch also changes that check to the faster exists method.

Change-Id: Ib6623fae5e31a81e032c3d429f6562cf2d13fb3e

cinder/volume/volume_types.py

@@ -70,11 +70,9 @@ def update(context, id, name, description, is_public=None):
     elevated = context if context.is_admin else context.elevated()
     old_volume_type = get_volume_type(elevated, id)
     try:
-        type_updated = db.volume_type_update(elevated,
-                                             id,
-                                             dict(name=name,
-                                                  description=description,
-                                                  is_public=is_public))
+        db.volume_type_update(elevated, id,
+                              dict(name=name, description=description,
+                                   is_public=is_public))
         # Rename resource in quota if volume type name is changed.
         if name:
             old_type_name = old_volume_type.get('name')
@@ -85,7 +83,6 @@ def update(context, id, name, description, is_public=None):
     except db_exc.DBError:
         LOG.exception(_LE('DB error:'))
         raise exception.VolumeTypeUpdateFailed(id=id)
-    return type_updated
 
 
 def destroy(context, id):