You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
It would be pretty awesome if memc could write to all upstreams on a cache miss, or support some function that will decide on a specific server based on crc of the key, this is how django works trough the python-memcached library example: python-memcached/memcache.py#L70
Right now a user will get as much caches misses as there are servers which is unfortunate.
An additional reason to support this is that memc currently will write to multiple upstream servers individually, but will not revoke the stale results on subsequent upstream writes, which may them result in stale cache hits, effectively making it impossible to use this lib with multiple upstream servers if stale hits are not acceptable.
somewhat related to #30 as this is a different solution ( maybe better ) for the same issue.
The text was updated successfully, but these errors were encountered:
helmus
changed the title
support writing to multiple upstreams
support writing to multiple upstream servers
Dec 8, 2017
helmus
changed the title
support writing to multiple upstream servers
support writing to multiple upstream servers ( sharding )
Dec 25, 2017
Uh oh!
There was an error while loading. Please reload this page.
It would be pretty awesome if memc could write to all upstreams on a cache miss, or support some function that will decide on a specific server based on crc of the key, this is how django works trough the python-memcached library example: python-memcached/memcache.py#L70
Right now a user will get as much caches misses as there are servers which is unfortunate.
An additional reason to support this is that memc currently will write to multiple upstream servers individually, but will not revoke the stale results on subsequent upstream writes, which may them result in stale cache hits, effectively making it impossible to use this lib with multiple upstream servers if stale hits are not acceptable.
somewhat related to #30 as this is a different solution ( maybe better ) for the same issue.
The text was updated successfully, but these errors were encountered: