i've not used NDB, but i use the GAE datastore quite a lot. lucky for me the projects i'm working on are OK with paying to use GAE so quota per-se is not a concern....but overall cost is.
some things that i have done: - i moved sessions to the DB, counter-intuitive i know, but they kept getting evicted from memcache and since google refuses to publish the memcache limitations i don't trust it for session data (at least not when i'm handling a sustained 30 request per second average 24 hours a day) - for "lookup" tables, or tables whose data changes very infrequently i have written module methods that read data from the cache if it exists, else read data the DB, convert it to a dict (Rows objects are not serializable and therefore not cacheable on GAE memcache) and store it in memcache. - i also make use of google edge cache by setting cache headers on responses that are static for some period of time. NDB may be interesting....but i've been too lazy to read up on it. :( cfh On Tuesday, March 5, 2013 4:49:32 PM UTC-8, Aleš Holubec wrote: > > I read about many limitations if I will use GAE for web2py, nevertheless > it will not stop me :-) > I think that main issue is to properly use memcaches. Is it enought to use > only cache.memcache and if yes what is the best way to use it? I plan to > use it for sessions a for selected results stored as dict, because I don't > want drain free quota on GAE to early. > But is it enough? Current DAL layer for GAE (google:datastore) is based on > db.Model but I think that some cache issues could be better solved if we > will use ndb.Model. I tried to modify dal.py to support it like google:ndb, > but I'm new in Python and I'm little bit shy to publish it because > modification is not so purely written and fully tested :-) > Do you have experience with that also? > -- --- You received this message because you are subscribed to the Google Groups "web2py-users" group. To unsubscribe from this group and stop receiving emails from it, send an email to web2py+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/groups/opt_out.
