Home / server / Questions / 215462
Accepted
Sam
Asked: 2010-12-23 12:10:26 +0800 CST

SQL cluster instance names for large project

  • 772

We're setting up two clusters. One dev and one prod. The Production will host two SQL instances - a OLTP and a DW.

The development will host 4 OLTP non-production environments and at least one DW non-production. We're working on getting more DW non-prods and possibly more OLTP systems.

I'm considering a naming scheme like this, where PROJ would be 3 initials for the project name.

Dev Cluster

  • MSSQLPROJD1\D1 (DEV)
  • MSSQLPROJD2\D2 (TEST)
  • MSSQLPROJD3\D3 (QA)
  • MSSQLPROJD4\D4 (STAGE)
  • MSSQLPROJD5\D5 (DW)

Prd Cluster

  • MSSQLPROJP1\P1 (PRD)
  • MSSQLPROJP2\P2 (DW)

To the left of the slash, each name must be unique network wide. On each server, the instance name, to the right of the slash, must be unique.

Any thoughts on this? I'm trying to avoid having instance names drifting from reality as the project progresses - say we change what we call a certain environment or want to repurpose one. Then we can update a listing of the purposes for the instances and be done with it.

How has a scheme like this worked out for you? Maybe you do things another way in your shop - tell me about it.

Thanks.

rev2

Dev Cluster

  • SQLERPD1\D1 (DEV)
  • SQLERPD2\D2 (TEST)
  • SQLERPD3\D3 (QA)
  • SQLERPD4\D4 (STAGE)
  • SQLERPD10\D10 (DWDev)
  • SQLERPD11\D11 (DWTest)*

Prd Cluster

  • SQLERPP1\P1 (PRD)
  • SQLERPP10\P10 (DW)

*hoped for, but not spec'd as of now.

sql-server sql-server-cluster

1 Answers

  1. Best Answer

    There are a million different naming standards which people use. There isn't really a right or wrong one to use, as long as the standard that you use works for you in your environment over the long term. The worst thing that you can need to do is to change your naming convention after you have picked one.

    Something to think about is how will this convention work if you add another Dev cluster, or another prod cluster. Will it continue to scale well?

    Personally I like to use a naming convention like this. You can easily modify this as needed with site names, etc as needed.

    Physical machines:

    SQL01A
SQL01B

    The Windows Cluster Name:

    SQL01

    The SQL virtual Names:

    SQL01V01
SQL01V02\INST1
SQL01V02\INST2

    This way you can quickly and easily see which physical machines a virtual name belongs two without having to log onto the server to look. And it scales nicely as you add another cluster that becomes what I've shown below. You can add more clusters easily, you can add more instances to any cluster without making things complex to figure out.

    Physical machines:

    SQL02A
SQL02B

    The Windows Cluster Name:

    SQL02

    The SQL virtual Names:

    SQL02V01
SQL02V02\INST1
SQL02V02\INST2
    • 3