Hi,

You raised an interesting issue I faced myself.

I'm not surprised by your results, I ran into the same conclusions some years 
ago.
0,3% is still too much from my point of view, but I think that your 
requirements are too high : I tend to consider that the time of a client 
redirection (HTTP 302) is more around 200 - 300ms (than 50ms). In this case, 
the rate of failures would decrease a lot and would be completely acceptable.

One more thing to have in mind is the way the synchronous EhCache replication 
works. It may have changed in the last version we use (2.7.0), but previously I 
remember that the synchronization with the other nodes was sequential which 
increase drastically the time required for replication if you have more than 2 
nodes.
The number of nodes in your CAS cluster is a big criteria in the choice of your 
storage system : for 2 nodes, Ehcache is a good solution. Starting from 3, 
maybe 4 nodes, I would prefer using Memcached.

Best regards,
Jérôme

-- 
You are currently subscribed to [email protected] as: 
[email protected]
To unsubscribe, change settings or access archives, see 
http://www.ja-sig.org/wiki/display/JSG/cas-user

Reply via email to