DB2 10.5 for Linux, UNIX, and Windows

Configuring a GDPC environment

Configuring a geographically dispersed DB2® pureScale® cluster (GDPC) environment involves getting the cluster installed and running, setting up GPFS™ replication, creating a database, and configuring the cluster for high availability.

Before you begin

Note: IBM support for a geographically dispersed DB2 pureScale cluster (GDPC) implementation requires engagement of IBM Lab Services for separately charged initial installation services. Contact your IBM sales representative for details.
Ensure that these prerequisites are met:
  • Three sites to communicate with each other through reliable TCP/IP links.
  • All DB2 pureScale Feature installation prerequisites have been satisfied across all hosts to be used in the cluster.
  • Two sites are connected via a WAN or dark fiber with distance range extenders, with a single high speed interconnect subnet configured across the sites.
  • The two sites each have a local SAN controller, and the SAN is zoned such that LUNs used for the DB2 pureScale instance are directly accessible from both sites. A one-to-one mapping between LUNs is required across sites so each LUN on the first site has a corresponding equally sized LUN on the second site.
Storage requirements on the three sites (Site A, B, and C):
  • Site A and B:
    • Provision equal sized LUNs on each site.
    • All LUNs are accessible by all hosts in both sites.
    • User data is replicated in these LUNs.
  • Site C:
    • The storage requirement is to assign a device (in /dev) to each shared file system in the cluster. In the following example where three shared file systems are intended to be created, three separate devices in /dev in Host T are required.
    • No user data is replicated to storage devices on Host T. These devices are used only to store replication metadata for recovery and maintain file system consistency purposes. As such, the size requirement for these devices is minimal (at least 50 MB.)
    • These devices do not need to be provisioned from the same SAN as in Site A and B. To keep the tiebreaker site as isolated as possible, use host local physical storage.
    • On AIX® operating systems, instead of hdisk as the device type, logical volumes can be used with the following guidelines:
      • Requires as many logical volumes as the total number of shared file systems in the cluster (each logical volume is at least 50 MB).
      • Create the logical volumes within in the same volume group.
      • Assign at least one physical hdisk to the volume group. The actual number depends on the number of logical volumes that are required, and the number of empty slots in the host for physical hard disk. If possible, for redundancy purposes, use two physical volumes.
      • If more than one physical volume is assigned, disable quorum check in the volume group.

About this task

For this example, the following hardware configurations are used:

There are three sites. In this example, the three sites are:
  • Site A: Hosts hostA1, hostA2, hostA3
  • Site B: Hosts hostB1, hostB2, hostB3
  • Site C: Host T
Equal sized LUNs have been provisioned on storage at 2 sites, and all LUNs are accessible by all hosts at the sites.
  • LUNs on disks located at Site A are:
    /dev/hdiskA1
    /dev/hdiskA2
    /dev/hdiskA3
    /dev/hdiskA4
    /dev/hdiskA5
    /dev/hdiskA6
    /dev/hdiskA7
Where /dev/hdiskA1 is used for the instance shared file system; /dev/hdiskA2 is used for the database log file system; and /dev/hdiskA3, /dev/hdiskA4, /dev/hdiskA5, /dev/hdiskA6, and /dev/hdiskA7 are used for the database data file system.
  • LUNs on disks located at Site B are:
    /dev/hdiskB1
    /dev/hdiskB2
    /dev/hdiskB3
    /dev/hdiskB4
    /dev/hdiskB5
    /dev/hdiskB6
    /dev/hdiskB7
Where /dev/hdiskB1 is used for the instance shared file system; /dev/hdiskB2 is used for the database log file system; and /dev/hdiskB3, /dev/hdiskB4, /dev/hdiskB5, /dev/hdiskB6, and /dev/hdiskB7 are used for the database data file system.
  • LUNs on disks located at Site C are as follows. These disks can be as small as 50 MB volumes. This example uses hdisk as the device type. On AIX operating systems, you can use logical volumes as the device type.
    /dev/hdiskC1
    /dev/hdiskC2
    /dev/hdiskC3
In this scenario, the geographically dispersed DB2 pureScale cluster (GDPC) is setup as:
  • Database MYDB is to be created on instance db2inst1.
  • db2inst1 has three file systems:
    • logfs for transaction logs and database metadata for MYDB.
    • datafs for database containers for MYDB.
    • db2fs1 for the shared file system for the instance
The command syntax in the examples use this format:
uid@host> command
Where uid is the user ID that executes the command, host is where the command should be executed, and command is the command to execute.

Procedure

  1. Install the cluster.
  2. Set up GPFS replication
  3. Configure the cluster for high availability
  4. Create the database.

What to do next

Once you have configured your GDPC environment, validate the GDPC through testing.