> From: Marvin Addison
> Sent: Monday, July 13, 2015 6:42 AM
> 
> I'd be curious to know why you don't like it. I encourage you to discuss 
> further
> on shib-users.

To be honest, I haven't really reviewed the memcached clustering implementation 
for the idp, my opinion is based on looking at the CAS memcached 
implementation, but I believe it is based on underlying memcached 
characteristics, not CAS specific implementation details, so is still valid.

The primary reason I did not like the memcached implementation was that it is 
not truly fault-tolerant. As I understand it, any given key is hashed and 
stored on one of the available memcached nodes. A client that needs it performs 
the same hash, and tries to retrieve it from the same node. I have a three node 
CAS cluster. With the memcached ticket registry, if one of the nodes were to 
crash, I would simply lose a third of my state. On the other hand, with the 
hazelcast ticket registry backend, a backup copy of the state on any given node 
is available on another node, and if you lose only a single node, you lose no 
state. Given my goal was to deploy a cluster not simply for load-balancing, but 
also for high-availability and fault tolerance, choosing a ticket registry 
backend that did not supply the latter was a no go.

> would be met by a single platform and it became a straightforward economic
> decision: consolidate and save. Of course our needs and yours are likely
> different; you'll have to evaluate for yourself whether a single platform 
> meets
> your requirements.

Thanks for the info. Barring the availability of an idp clustering backend more 
like Hazelcast in terms of fault tolerance (perhaps somebody at Unicon might be 
interested in putting one together :)? ) I think I'm leaning towards keeping 
them separate with the cluster state on the CAS side for now. Thanks much for 
the information…


-- 
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