[cas-dev] high availability issue of CAS

[Lin George]

Thanks Robert,

I read your referred document, and it is really nice! But dues to my limited English skills, I still have one confusion which is 100% clear after I read it.

Suppose we have two machines to make a cluster, say machine A and machine B, suppose user P's session information (stateful information CAS used) is stored in machine A, and machine B stores user Q's session information.

If machine A is down, what will happen to user P? Must user P must enter again his/her credential and using machine B to authenticate? Or we could use the so-called cache replication approach to store P's credential information into machine Q?

regards,
George

See http://www.ja-sig.org/wiki/display/CASUM/Clustering+CAS

Robert

Am 18.11.2008 um 13:35 schrieb Lin George:

> Sorry for sending to multiple list because I am a new member and do  
> not know which list is better to send. It is appreciated if anyone  
> could advise me which list is the correct one to send this question.  
> I am a CAS new user not developer who wants to contribute CAS code  
> (I may contribute after I am familiar with CAS). :-)
>
> My question is, after reviewing CAS architecture, I have not found  
> how CAS ensures high availability and scalability.
>
> More specific question like this,
>
> - if I only setup one CAS authentication server (service), and if it  
> is not availabile for some reason (e.g. power-off), how to ensure  
> the whole authentication system (including web client and web  
> resources) still works? Or the whole system will not work -- i.e.  
> without authentication server (service), even if end user provides  
> valid credential, he/she can not access the web resources?
>
> - if the user base is large, how to setup multiple CAS  
> authentication servers (service) to balance load? I did not find  
> related information in the CAS wiki page.
>
> Any advice are welcome.
>
> thanks in advance,
> George