Hi Ikai
see the a href=http://i52.tinypic.com/28hndxk.jpg;
target=_blankapp stats/a
@905ms datastore_v3.RunQuery real=1090ms api=8225ms
@2504ms datastore_v3.Next real=339ms api=4711ms
my q.fetch(2000) method takes ~3.5 seconds
On Mar 23, 12:42 am, Ikai Lan (Google) ika...@google.com wrote:
I
Hi Ikai, is there a measure to say the entities are large? is it about
the number of
columns or blob or text? if so in my entities I have string and number
properties only.
Yes. I need all the 2000 (some times bit more) entities at once for
reporting.
Currently I'm not seeing a way where I can
I think you should run AppStats and see what's going on. 2000 entities is
never going to be particularly fast, but 3 seconds might be a bit on the
high side. Run the tool to be sure.
Ikai Lan
Developer Programs Engineer, Google App Engine
Blog: http://googleappengine.blogspot.com
Twitter:
q = db.Query(PrimaryData)
q.filter('SheetMetadataId', metadata.getSheetId())
return q.fetch(2000)
Hi Ikai, even the above query takes 3.3 ~ 3.6 seconds. The number of
entities returned by the query is 1261.
I'm using Expando model, the total number columns for this particular
set of entities are
If that's the performance, then I'm inclined to say the entities are large
and that's about the performance you can expect. One more thing to keep in
mind: the cost of deserializing each entity is non-trivial.
Is there a reason you need all 2000 entities? Perhaps there is a different
way to
Hi Robert,
I've not used reference properties. Its a simple query and it'll be
like this.
q = db.Query(PrimaryData)
q.filter('SheetMetadataId', metadata.getSheetId())
q.filter('InstanceAssignedTo IN',
[u'User_c42e8919_448e_11e0_b87b_f58d20c6e2c3',
u'User_1fd87ac5_073d_11e0_8ba1_c122a5867c4a'])
If you have a composite index, the performance of this query should be
mostly linear (find start location of index and just return min(2000, number
of entities that satisfy query from there on out) number of keys and items.
If you only have individual indexes for SheetMetadataId as well as
