Questions on High Availability using idle standby
I am working on implementing RTC server High Availability using idle standby following instructions posted at http://jazz.net/library/article/207 and have some questions to ask by hoping someone has tried it before.
1. Do I need two physical server boxes for the primary and backup servers or I just need two instances of WAS servers created using profile conception on the same box?
2. If both is OK, what is the recommendation and advantage to use one instead of other?
1. Do I need two physical server boxes for the primary and backup servers or I just need two instances of WAS servers created using profile conception on the same box?
2. If both is OK, what is the recommendation and advantage to use one instead of other?
3 answers
I finally got some answers from RTC support and think you might also be interested:
I just confirmed with the author of this technote https://jazznet/library/article/207, you will need two physical machines with one WAS profile on each. The idea is to ensure if one server goes down, the other one can act as a backup. If you put them into one single machine, with two different WAS profiles, if the machine is down, for example, due to hardware issue, then the other RTC instance can't act as a backup since it is hosting in the same physical box.
And there is more:
Yes, you need give the backup server a unique name and static ip address as well. The unique host name will need to be included to the http server's plugin.xml.
No, users will not have to modify their connection, because the http server is configured to route access requests to either the primary or backup application server, using the connection information stored in the plugin.xml. Your users will be accessing jazz through the http server at the front end, like instead of using http://RTChostname:9443/Jazz, they will be using http://httphostname/jazz, so they don't need to upate anything from their end.
If the http server is down, you can instruct your users to access Jazz directly to bypass the http host, like using http://RTChostname:9443/Jazz instead of http://httphostname/jazz. You could also look for ways to have a backup server for the IBM HTTP server. Deployment details will need to be discussed with the HTTP server support team.
As we know now we need to use two separate VM boxes to be the Primary Server A or Backup Server B, and we've had primary server up running with a unique name and static IP address, do I need to give the backup server a unique name and static IP address too?
Yes, you need give the backup server a unique name and static ip address as well. The unique host name will need to be included to the http server's plugin.xml.
If yes, then when primary server is down, we faileover to the backup server, all users will have to modify their connection url to point to the new server name and IP address. Is this worked as being designed?
No, users will not have to modify their connection, because the http server is configured to route access requests to either the primary or backup application server, using the connection information stored in the plugin.xml. Your users will be accessing jazz through the http server at the front end, like instead of using http://RTChostname:9443/Jazz, they will be using http://httphostname/jazz, so they don't need to upate anything from their end.
What happens when the IBM HTTP server between the RTC end users and RTC servers (primary and backup) itself is down? Will the RTC end users still be able to connect to RTC primary or backup server while we just failovered?
If the http server is down, you can instruct your users to access Jazz directly to bypass the http host, like using http://RTChostname:9443/Jazz instead of http://httphostname/jazz. You could also look for ways to have a backup server for the IBM HTTP server. Deployment details will need to be discussed with the HTTP server support team.