Last week i had an issue with Sharepoint. The customer i’m working for, wanted to enable my sites for all users (+2500).
I was afraid that 1 WFE would be insufficient and critical.
The problem with mysites is, that every user is sitecollection administrato for his own mysite (every mysite is a site colelction)
Because i didn’t want the preformance to degrade or the main portaal te be inaccesible when users start “experimenting” with their new functionality, i adviced adding an extra WFE-server.
Although a lot information is found why you should ad a WFE server, i couldn’t find any information about how to implement an extra WFE server.
After some digging, testing and discusion on forums, i found out there are 3 things you need to know before installing an extra WFE
- Service level of your farm.
- Level of redundancy you need
- non custom customizations
1 service level
we already updated our current farm with SP1 and infrastructure updates (minimal level as of january 2009)
Although Sharepoint with sp1 integrated is around, we still needed to install the infrstructure updates prior to joining the farm.
The service level of the computer joining the farm and the service level of the farm need to match, or alse you;ll get an error.
this was my first problem: how to update without having a working installation.
This was easier then expected: simple install sharepoint, but don’t configure yet. Then install al updates, but again, don’t configure. You even might get errors opening Central Administration. Simply ignore them. When all updates are installed, start the configuration. Then add your server to the farm. This might take some time, because all web application and all content is copied.
Now we have 2 WFE -server. They should be able to show you the Sharepoint content simply by going to there hostnames in IE.
Example: http://wfe1 and http://wfe2 should both give the same page.
In out case, the second didn’t work correctly. After some investigating, it seemed that the web.config wasn’t copied correctly.
After copying it manually> everything works.
2 Level of redundancy you need
Although we have 2 WFE servers, we don;t have any redundancy. People need to choose one of the servers names to get there.
we would rathet have 1 name that is an alias for both.
This can be achieved in 2 ways
A use round robin
In this config you use a single name (in DNS) that point to the 2 servernames. Depending on when people as access the DNS-server servers one of the names. This is some form of redundancy. If one goes down, the dns server can be configured to only serve the working server.
B use network load balancing
In this config both servers are configured in an NLB configuration. This is standard software of windows. You;ll need to change the configuration of the windows servers. The servers are combined to a new NLB-server with it’s own ip-adress. This ip-adress is listed in the DNS-server as the ipadres of the Sharepoint portal. When ou’ll reach that ipadress, you reach the NLB-server. This server will then use it’s configuration to check which server is least busy and if it’s working. So, if not both servers are down, you will always reach the least busy server. No change to the DNS server is needed.
You might need to adjust you AAM because it needs to listen to a second server.
3 non custom customizations
Every customization that’s done without using features, is NOT deployed to the second server. You need to do these customizations by hand. This is also the reason why you shouldn’t make changes to the system without using features.
We now have a working environment with 2 WFE servers (we used round robin)
Eric van Glabbeek