Mark,
Thanks very much for this - it's useful. I'm slowly getting used to
this non-relational way of thinking.
The count of the tags doesn't need to be 100% accurate, as it will
only be used to approximate how many entries exist for a given tag -
the tag cloud aim is to show tags associated with many entities in a
larger font and tags associated with fewer entities in a smaller
font. If the insert, update or delete of Entity succeeds and gets
committed, but the persistence action against the TagCounter fails in
a few instances that would not be a big deal.
I'll probably allow three tags per Entity, so was planning on:
class Entity {
List<String> tags;
String foo;
}
for flexibility in of the number of tags.
So when inserting a new entity I would need to:
Create Entity
For each tag in Entity select TagCounter
If tag counter exists for tag increment count by 1
If tag counter does not exist for tag create TagCounter with count of
1
When deleting an entity I would need to:
For each tag in Entity select TagCounter
Decrease count by 1 of TagCounter
Delete Entity
This sounds ok to me. When a user clicks a tag the query is simple
too.
I'll give this a try and see how it looks.
Thanks again.
On Jul 21, 12:46 pm, Mark <[email protected]> wrote:
> I probably have a naive understanding of what you're doing, maybe
> explain a little more, but to me it looks like this:
>
> class TagCounter {
> int count;
> String label; // primary key, unique tag name.
> }
>
> class Entity {
> String tag;
> String foo;
> }
>
> whenever the user creates a new entity, record its tag(s), then
> increment the appropriate TagCounter instance. Keeping counters like
> this introduces a bottleneck and goes against what you want to be
> doing for a highly scalable web service, so you may want to shard them
> (imagine you have thousands of users all trying to increment a single
> tag counter at once). Since the above two classes won't be in the
> same entity group, you wouldn't be able to perform both operations in
> a single transaction. You can either accept the fact that your tag
> count may be out of sync with the actual number of entities with that
> tag, or you can have a multistep process and clean up dangling tag
> counts with a task (eventual consistency might be the term here). I'm
> referring to Nick Johnson's article on distributed transactions which
> I think you can modify to do what you're looking for:
>
> http://blog.notdot.net/2009/9/Distributed-Transactions-on-App-Engine
>
> I'm still learning app engine so take the above with a big grain of
> salt. To me, it seems that if you want to write a highly scalable web
> service, you need to accept that you may lose some data precision, or
> you have to put in additional work to get what you want (compared to
> assuming a single mysql instance etc).
>
> Anyway, after you implement something like the above, your original
> two queries should be easy to do. You probably want to define how many
> tags an Entity can have. If an Entity can have more than one tag, I'd
> explicitly state what that max is. Write out everything you need when
> the user submits an entity, so that your reads will be fast as other
> users browse the site.
>
> Mark
>
> On Jul 19, 9:16 am, planetjones <[email protected]> wrote:
>
>
>
> > Hi,
> > I'm looking to store entities using GAE Java which have 1-many tags.
> > I
> > would like to display a tag cloud, so will need to know how many
> > times
> > each tag occurs (can't use aggregate functions and group by on GAE to
> > do this like I would in SQL). And when I click a tag I would like to
> > retrieve all entities with the selected tag.
> > Does anyone have any examples of what my Classes should look like to
> > do this optimally and efficiently? Sample JPA code would be great
> > too.
> > Cheers - Jonathan
--
You received this message because you are subscribed to the Google Groups
"Google App Engine for Java" group.
To post to this group, send email to [email protected].
To unsubscribe from this group, send email to
[email protected].
For more options, visit this group at
http://groups.google.com/group/google-appengine-java?hl=en.
