2 Iron

DataDomain ネットワーク統計コマンド各項目の意味



system show performance

replicatoin show performance all



engineering wikiに情報があるようなナレッジを見つけましたが、



Data culled from the engineering wiki at http://iweb.datadomain.com/eweb/pmwiki/pmwiki.php/DataDomain/SystemShowPerformanceCommand


0 Kudos
3 Replies

Re: DataDomain ネットワーク統計コマンド各項目の意味


Please see the following KB articles:

  • System Show Performance Background Activities 180735
  • System Show Performance Cache Miss 181670
  • System Show Performance Compression 181667
  • System Show Performance Maximum Net Throughput 181663
  • System Show Performance Utilization (proc, recv, send, idle) 180743
  • System Show Performance Utilization (cpu & disk) 181672
2 Iron

Re: DataDomain ネットワーク統計コマンド各項目の意味

Thank you for KB articles,

Do you have meaning of colums for "replication show performance all" too ?


0 Kudos
2 Iron

Re: DataDomain ネットワーク統計コマンド各項目の意味

"replication show performance all" の各項目ですが、英文ですがCommand Reference Guideに記載がございますので抜粋し貼付させて頂きます。

replication show performance {obj-spec-list | all} [interval sec]

[count count]

Display current replication activity. Default interval is two seconds. All users may run this command option.

If a single source context is specified, four additional columns are presented. These columns show the relative amounts of time spent working on, or waiting for, replication sender threads for the specified context. Values are calculated by the amount of time spent for the activity, multiplying the time by 100, and dividing the time by the duration of the reporting interval. Values can exceed 100 due to the presence of multiple threads working on behalf of the specified replication context.

When a replication throttle is configured, users may see a large amount of output followed by a period of none while viewing performance or statistics. This behavior is the result of how statistics are calculated by replication, combined with the default Data Domain system replication configuration of using large TCP buffers. When a throttle is in effect, data is buffered before being sent on the network.

Users may configure the replication bandwidth and delay arguments in the replication option reset command on the source and destination to use smaller TCP socket buffers.

This reduces the total amount of data on the network, increases how often replication writes data to its sockets and, as a result, increases the frequency of updates for statistics counters.


Output Definitions


Pre-comp (KB/s)

The size value before compression is applied. Sometimes referred to as logical size.

Network (KB/s)

The amount of compressed data transferred over the network per second.


An internal system resource associated with reads and writes. One replication context can use multiple streams for better performance.


The time spent reading filesystem data from the local filesystem. Typically this number is the second highest number after Network. On a deployment with high network bandwidth, Reading may be the largest column.


The time spent on miscellaneous bookkeeping activities and replicating filesystem namespace operations. Typically this value is under 50. If this value exceeds 50 on a sustained basis, it may indicate an unusual workload (a large number of file

attribute updates, for example).


The time spent waiting due to the receiver not providing the sender enough information on what data to send. Typically this value is low. Exceptions include systems on high-speed networks where the sender is a more powerful Data Domain

system than the replica, or where the replica has a higher workload than the sender because the replica is the destination for a multiple replication contexts.


The time spent sending file data and metadata and waiting for replies from the server on what data needs to be sent. Typically this is the highest of the four values.

This value exceeds 100 regularly if the sender is able to replicate multiple files in parallel.