Ian Cooper wrote:
A "cache" meaning? We had this definition discussion years ago in WREC...
So let's take advantage of the WREC work. It should be a 'caching
proxy' then...
Yes. But when a single caching proxy became a bottleneck we found other
ways of clustering multiple caching proxies to fix that problem...
... but we now have additional functionality/data on these 'caching
proxies' (i.e. the OPES intermediaries), for example rule processors
and rule modules. Clustering multiple of such extended 'caching
proxies' requires additional consideration, for example keeping the
rule sets consistent etc. It would also imply that you duplicate the
rule engines with clustering the 'caching proxies', while this would
NOT be necessary of you have one OPES intermediary serving a cluster
of remote callout servers...
guess the issue to be considered is whether one option scales better
than the other (and perhaps in which cases).
Agreed - and this fit also fit in the "Scenario Document".
-Markus