Yesterday I came across this gem of a Blog post on Nick Johnson’s Blog about how to go about efficiently storing Google App Engine datastore objects in Memcache on the cloud.

I had previously read that the Datastore models were large and carried around a lot of baggage (software baggage that is, not the stuff on wheels!) – this is one reason why you are discouraged from passing them as arguments to tasks on task-queues.

This blog shows you how to extract and store only the essential info from the objects for storage in memcache using th methids model_to_protobuf () and model_from_protobuf()

Anyway, enough chat – here’s the post:

http://blog.notdot.net/2009/9/Efficient-model-memcaching

1. Peter K. Says:
   July 29th, 2011 at 9:11 pm

   Yes. We used Google protocol buffer encoding of data on the embedded access control system I worked on. It’s a very efficient way of encoded data in a binary format. We looked at using SOAP, but that required about 100 times the bandwidth.