A successful implementation of node replication relies on sufficient, dedicated hardware resources. Increased amounts of memory and processor cores are required. The database and its logs must be appropriately sized to ensure that transactions can complete. A dedicated network, with enough bandwidth to handle the amount of data you intend to replicate, is required.
Use the checklist to verify that hardware and your Tivoli® Storage Manager configuration have characteristics that are key to good performance.
| Question | Tasks, characteristics, options, or settings | More information |
|---|---|---|
| Are you using high-performance disk for the Tivoli Storage Manager database? | Ensure that the disks that are configured
for the Tivoli Storage
Manager database
have a minimum capability of 3,000 I/O operations per second (IOPS).
For each TB of data that is backed up daily (before data deduplication),
add 1,000 IOPS to this minimum. For example, a Tivoli Storage
Manager server that
is ingesting 3 TB of data per day would need 6,000 IOPS for the database
disks:
|
Checklist for server database disks |
| Are you using enough processor cores and memory for node replication and optionally, data deduplication? | If you are using node replication without
deduplication, use a minimum of 4 processor cores and 64 GB of RAM
for both the source and the target servers. For any server that is configured for node replication and data deduplication, use a minimum of 8 processor cores and 128 GB of RAM. |
|
| Have you properly sized your disk space for the database, logs, and storage pools? | To determine whether your database can handle
the additional space requirements, you must first estimate how much
more database space node replication uses. For the active log, use a minimum size of 64 GB for node replication. Use the maximum allowed size for the active log, which is 128 GB if you are also using data deduplication. Make the archive log space at least as large as the space defined for the active log. Also, specify a directory for the archive failover log in case it is needed. |
See Determining server database requirements for node replication. |
| Is your network capable of handling the additional traffic for the amount of data that you intend to replicate between source and target servers? | For node replication, network bandwidth
must be greater than the planned maximum throughput. You can estimate network bandwidth that is based on the amount of data that you are replicating. |
Estimating network bandwidth for node replication |
| Are you using data deduplication with node replication? | By using data deduplication with node replication, you reduce the bandwidth that is required for replication operations. Data deduplication reduces the amount of data that is sent to the target of the replication operation. | Measuring effects of data deduplication on node replication processing |
| Have you scheduled node replication in the optimum order for the daily schedule? | Ensure that you are running node replication
after client backup. If you are using data deduplication for the replication process, run identify duplicates to completion in the primary storage pools before you run replication. |
For more information, see the following topics: |
| Have you optimized the number of sessions that are used for sending data to the target replication server? | You can improve replication performance
by using the MAXSESSIONS parameter on the REPLICATE
NODE command to specify data sessions. The number of sessions that are used for replication depends on the amount of data that you are replicating. |
Managing the number of replication sessions |
| Do you have enough mount points to avoid stalled replication servers and other server processes? | Determine the number of logical and physical
drives that can be dedicated to the replication process. For example,
if a library has ten tape drives and four of the drives are used for
another task, there are six available drives for node replication. Specify the number of mount points you require and ensure that there are drives available to complete node replication. |
Typically tape is not used for node replication except for the initial replication. |
| Do the node replication processes complete replicating all newly ingested data before the beginning of the next backup cycle? | If replication processes cannot finish before
the start of the next backup cycle, consider the following actions:
|
|
| If you are using data deduplication with node replication, do the processes for identifying duplicates complete before the start of node replication processing so that data deduplication is used to its full advantage? | If the process completes, or goes into an idle state before node replication begins, then all new data is being processed. |