opencog-dev team mailing list archive

["David Hart" <hart@xxxxxxxxxxxx>]

Hi All,

An interesting bit about scaling the LinkedIn database:

http://perspectives.mvdirona.com/2008/06/08/ScalingLinkedIn.aspx

So, if I understand correctly, each of the 40 replicated instances of the DB
(the entire social graph of LinkedIn) is indexed differently, and handles
different queries, depending on query type, as a form of load-balancing.

It seems to me that OpenCog's goal of dynamically building and optimizing
indexes is a better approach, and scaling for load then becomes a problem of
replication and synchronization across instances (and also depends on query
type, as some types of queries will not need a completely fresh replica).

-dave