Sorry if i am being too dumb. I got into a confusion after trying to understand the environment how its setup here. Users are in two different countries and they are maintaining two independent BO servers, each universe and report is doubled ( one in each server). When i tried to find out the reason the answer i have got is because of performance issue. If server is located in country A then accessing reports in country A is faster but slower when trying to run them from country B. Somehow i cant seem to believe it. My understanding is that when both servers are accessing same database the it shouldnt be a big issue on the performnace side. I have worked in environments where users are across countries and yet did have only one setup for BO. Am i missing something ?? I am more confused now.
Am i right in telling them that shouldnt cause performance issues ? Or can i suggest any better plan so that they dont have to duplicate the work and maintain 2 sets of each ? Clustering servers is possible or helpful ?
Please someone can help me straighten this in my mind ?
There are three tiers to consider – the user, BO server, and database server. Generally you will notice better network performance if the nodes are close together, so it’s reasonable that putting a separate BO cluster close to the users would help performance.
If users are refreshing reports interactively, then there still would be a connection from the BO server to the database. However, high latency would not be as obvious of an impact to the users.
Having two entirely separate clusters is a reasonable way to ensure processing occurs close to the users, but there are other methods as well. You can create servers and server groups that force processing to be done on a specific node. There would be one cluster with a node in each country, but the web app and WebI processing would occur in the node closest to each user.
You can also use Federation to sync content between two clusters. You still have to maintain two individual clusters, but universes & reports can be automatically synced. You can even force a scheduled report to be processed on the origin site if it’s created from the remote cluster.
It will matter where the target reporting server is. Odds are that it is located in the country with the better report performance because that’s where the bulk of the data transfer will happen - in getting the query results back to Webi.
(BOB member since 2008-06-06)
(BOB member since 2002-08-29)