SQL Server 2000 & 2005 Clustering
Chapter 6 – Designing your SQL Server Cluster
Preparing Cluster Resource Groups
As a first step, the DBA should list all server-based applications and services that will run within the cluster environment, regardless of whether they will need high availability. Applications and services that need to be highly available should be placed into resource groups. Other applications should be tracked, and their interactions with clustered applications and services should be clearly understood. Failure of an application or service that is not part of a resource group should not impact the core functions of the solution being offered. If it does, the application or service may need to be clustered. Failover should be initiated only by critical resource or services. Non-critical resource or services should not initiate the failover.
Preparing for the SQL Server Cluster Installation
Verify that your operating system is installed properly on both cluster nodes and it is up and running. Scroll through the event log and make sure there are no operating system-related errors. Operating system can be any among Windows 2000 Advanced Server, Windows Datacenter Server, Windows 2003 Enterprise Edition, or Windows 2003 Datacenter Server.
Clustering Services Installed on Local Hard-Drive
Verify that clustering services are installed on local hard drive of the both physical nodes and that it is working fine. Examine the event log and make sure that there are no cluster services-related errors. You can verify the cluster health by opening the cluster admin.
Memory Switches Configured
If your Node Memory is 4 Gigs or more, make sure that memory switches are configured properly for your operating system. For more details on memory switches, visit the Best Practices section of this article.
Network Connection Between the Two Physical Nodes
Verify that both nodes are in the same domain, and that both nodes are accessible from each other. As per Microsoft documentation, an important factor here is that network latency from one node to another should not exceed 500ms.
Ensure that each node has a local storage drive available consisting of at least 2 Gigs of free space. This space will be used for placing the SQL server binary files.
Cluster Administrator Account
This account is required to perform the administration activities related to cluster installation, maintenance, and failover. This account must be configured at the domain level with administrator privileges on all nodes within the cluster group. A secure and complex password should be used for this account.
All Shared Disks Visible From All Nodes
Make sure that all shared disks or SAN drives are visible from all the nodes in the cluster.
Verify that this drive is accessible from all of the nodes in the cluster. In case of Active/Passive cluster setup, the Active node must be defined as the preferred owner of this disk resource. Typically, this quorum drive is assigned with drive letter “Q:”.
Drive Letters on Each Node
Check to see that drive letters on each node are same when connected to the shared array.
Clustering Failover Tested
It is advisable to test the failover process before you start the SQL Server cluster installation. There are couple of ways to do this.
1. Do a move group with cluster admin.
2. Bring down the cluster nodes one by one.
As a part of cluster design, prepare the names which you would like to assign to different components of the cluster architecture. It is preferred that you come up with a combined component name based on a meaningful combination of different business aspects.
Here are few examples.
Business Unit Name MS, GM, BMW
Server category PRD, TST, DEV, INT, ACP
Application Name SLS, MRKT, MFCT
Geographical Site Name EUR, USA, AUS, JPN
Server Type WIN, CLS, SQL
Physical Node GM TST SLS AUS 001 WIN
Virtual cluster GM TST SLS AUS 001 CLS
SQL Named Instance GM TST SLS AUS 001 WIN SALES