App Engine's memcache service provides the ability to get multiple keys in one API call (get_multi(['1', '2'])), and although other answers here have shown that there is no hard limit imposed in memcache (the example I saw tested out getting 100,000 keys in one call), obviously there are practical limits on App Engine as the number of keys grows and latency becomes unreasonable.
Has anyone run any benchmarks showing how long it takes to retrieve multiple keys as the number of keys grows into the thousands?
