We are having a bit of a debate in the office with regards to clustering, named instances and default ports.
While I am convinced I am correct, I am often happy to be disproved and I'm wondering of anybody out there has experience with this.
We have an existing SQL Server 2008 R2 cluster (on Windows 2008 R2 server). We have several services on there presently. For arguements sake we have the following config:
Cluster name
CLUSTER01 : 192.168.1.1
Services/Instances:
INST1\SQL1 : 192.168.1.2,1502 INST2\SQL2 : 192.168.1.3,1503 INST3\SQL3 : 192.168.1.4,1504
So, within DNS, there are entries for INST1, INST2, INST3 and they have the IP's assigned accordingly. Users can then connect using DNSNAME\INSTANCENAME or DNSNAME,PORTNUMBER and so on and so forth...
We want to add a forth instance, INST4\SQL4 (IP will be 192.168.1.5 and port will be decided in a little) but here's where we get a bit more complicated... we want to add a SECOND DNS entry for INST4 called DB1. This will also point to 192.168.1.5 but we do not want to pass in a port number to access the database.
So, it was proposed to me that if we leave the SQL Default port of 1433 for this instance, SQL browser will simply see the incoming request to 192.168.1.5 and redirect to the instance assigning the port 1433 tot he end and Bob's your uncle... there was also mentions of a server side alias to assist in this.
Personally, I dont think that this will work at all, I think that aliases only work client side, and the default port is all well and good, but it will only point to the default instance, and not a named instance with port 1433...
Any advice or experience with this would be welcomed! We do not want to install SQL clients locally on all user machines and have client side alises as this is too difficult to maintain.
Regards,
Andy