Could I extend this topic a little further?   Imagine a situation in
which there is redundant web service feeding a metakit database.
Either because of load sharing among servers (which are geographically
distributed) or because of fallback on failure of the principal
server, the databases diverge on the various servers.   What is
wanted is for all the databases to update mutually, so as to present
an apparently seamless service where the end user neither knows nor
needs to know which server he has been using.

This is prompted by recent experience when I discovered that one of
our ISPs had our colocated machines in a building without adequate UPS
provision, and we suffered a nearly 24-hour outage.   No real damage
done that time, but we have to move to a more redundant situation.

Has anyone tackled such a requirement successfully?   We're only
talking maybe 500 database updates per day, but the service needs to
be in place continuously, and preferably without manual intervention.
 
_____________________________________________
Metakit mailing list  -  [EMAIL PROTECTED]
http://www.equi4.com/mailman/listinfo/metakit

Reply via email to