Targeted SLP
A large number of A.I.R. Replication jobs running concurrently can cause performance problems due to contention. Limiting the number of concurrent A.I.R. Replications can allow each of the active replication jobs to perform at their maximum potential given the underlying environment. At NetBackup 8.3, it is possible to limit the number of concurrent A.I.R. Replication jobs by setting a SLP.REPLICATION_TARGET_JOB_LIMIT value. This value can be applied globally, affecting all A.I.R. Replications, or by target storage server. This setting limits the number of A.I.R. Replications going to a target storage server.
SLP.REPLICATION_TARGET_JOB_LIMIT = <limit_spec>[,<limit_spec>][,...]
It is possible to throttle replication jobs globally or per target server. Each limit-spec value can be a numeric value which sets the limit for every target storage server. It can also take the form <storage_server_name>:<number> which sets the limit for a specific target storage server. The two limit types can be specified on the same parameter setting, as shown in the following examples. (First enter nbsetconfig from a command line. Then enter the SLP.REPLICATION_TARGET_JOB_LIMIT values.)
# Set the replication limit for each target storage server to 10. ->nbsetconfig nbsetconfig>SLP.REPLICATION_TARGET_JOB_LIMIT = 10 nbsetconfig><end-file marker - Unix: Ctrl+D Enter, Windows: Ctrl+Z Enter> # Set the limit for two named target storage servers to 12 and 6. Set the limit for # all other target storage servers to 8. ->nbsetconfig nbsetconfig>SLP.REPLICATION_TARGET_JOB_LIMIT= targetServerA:12, targetServerB:6, 8 nbsetconfig><end-file marker - Unix: Ctrl+D Enter, Windows: Ctrl+Z Enter>
The advantage of Target A.I.R. is that it helps to gain more granular control over the number of Replication jobs that will run at one time to specific storage servers acting as targets. Prior to this, the only option was to greatly increase the Maximum size per A.I.R. replication job to every increasing levels, but this would not set a specific number of running operations. Also, different target storage servers were all treated the same, no considerations for the capabilities of different storage servers were possible. By adding the SLP.REPLICATION_TARGET_JOB_LIMIT parameter, limits can be set on the number of A.I.R. Replications that can run per target, while still keeping the Maximum size per A.I.R. replication job parameters in place to match needs. Also, specific limits can be put on specified storage servers to accommodate the capabilities and conditions of different storage servers.
The SLP.REPLICATION_TARGET_JOB_LIMIT parameter is limited to A.I.R. Replications. More information is available in the following technical article:
How to tune NetBackup Auto Image Replication (A.I.R.) operations for maximum performance
Replicating small images can create delays between the replication of each image. Between each image there is a check done to make sure the image is replicated. With small images this check can be longer than the time needed to actually replicate the image which can add up to a significant time. It is possible to adjust this delay for checking the image using AIR_POLL_INTERVAL_TIME setting. The size of the image to check can also be adjusted using the AIR_POLL_INTERVAL_CHUNK_SIZE setting.
AIR_POLL_INTERVAL_TIME = <seconds>
This option sets the time factor to a fixed number of seconds and should be used when it's desirable to optimize the time needed to complete the replication. It will generate the most overhead. Use this option with an initial value of 1 to minimize the time needed and increase the throughput. Increase the value if the overhead generated impacts system performance.
AIR_POLL_INTERVAL_CHUNK_SIZE = <size_value>
This option attempts to calculate the best time factor based on image size. The goal is to set the time factor to approximate the time it will take to replicate a typical image. The size value should be a decimal number representing the number of bytes that can be replicated in 1 second. The default value is 104857600 (100 megabytes). Increasing the chunk size value will decrease the time factor and improve the accuracy of the completion detection at the cost of increased overhead.