Manual Pages


Table of Contents

NAME

awa - Control the Automated Workload Analyzer (AWA)

SYNOPSIS

awa start aggr

awa print [ aggr ] [ -l intervals ]

awa stop [ aggr ] [ -q ]

DESCRIPTION

The awa command controls the Automated Workload Analyzer (AWA), a Data ONTAP capability that estimates the Flash Pool cache size and estimates the cache hit rates across reads and writes.

AWA can be enabled on any Flash Pool or HDD aggregate. It monitors the throughput and access patterns for the workload happening on an aggregate. By analyzing the statistics collected, it summarizes the key characters of the workload. It also models Flash Pool cache to estimate the potential effect of implementing a Flash Pool aggregate with various cache sizes using the default caching policies.

Aggregates
The AWA modeling does not take the effect of existing Flash Pool cache into consideration so it is capable of estimating the potential benefit of a larger SSD tier size.

Intervals
AWA employees a coarse-grained time unit as the workload statistics measuring unit as well as the precision of statistics maintained. AWA maintains statistics for each interval for a history of up to 1000 of them. The interval is 10 minutes. The values from the summary output of AWA are based on the maximum values observed during each interval considered.

Summary Outputs
The AWA CLI command awa print generates the summary output. AWA Summary Output is a set of statistics that captures the key characters of the workload being observed. There are two tables presented in the Summary Output. The first one is the Workload Character Table which presents the general workload statistics. The other one is the Cache Size Effect Table which captures hit rates for various cache sizes. Below are the descriptions for each field.

Workload Character Table
Summary of the past `n` intervals:

Read Throughput
Read throughput from the storage disks

Write Throughput
Write throughput to the storage disks

Cacheable Read
Non-sequential reads: percentage of data reads that could have been cached

Cacheable Write
Random overwrites: percentage of disk writes that could have been cached

Projected Cache Size
Projected Flash Pool cache usage

Projected Read Offload
Percentage of blocks that could be read from the Flash Pool cache instead of HDDs

Projected Write Offload
Percentage of block overwrites that could go to the Flash Pool cache instead of HDDs

Cache Size Effect Table
Summary Cache Hit Rate vs. Cache Size

Size
Percentage of the "Projected Cache Size". The cache sizes are in 20% increments.

Read Hit
Estimated read hit rate when cache size is a fraction of the optimal one.

Write Hit
Estimated write hit rate when cache size is a fraction of the optimal one.

USAGE

awa start aggr
awa start enables awa on the specified aggregate. Resources are allocated and statistics start getting collected.

awa print [ aggr ] [ -l intervals]
awa print outputs the statistics summary of the running awa on the specified aggregate. This command by default summarizes the entire monitoring period up to AWA's monitoring limit of 1024 intervals. When the number of intervals is being supplied by the -l option, only that number of intervals into the past will be covered by the summary. (To have the AWA readings of any specific interval, this command needs to be supplied with -l 1 periodically at the end of each interval.)

awa stop [ aggr ] [ -q ]
awa stop stops the awa and generates the summary output, then releases all of the resources. The -q option suppresses the default summary output when stopping AWA.

EXAMPLES

Sample Summary Output
  ### FP AWA Stats ###

                       Host lyra22a                    Memory 16706 MB
              ONTAP Version NetApp Release R8_2_1x Cluster-Mode
                AWA Version 1
             Layout Version 1
                 CM Version 1

  Basic Information

                  Aggregate aggr1
               Current-time Wed Nov 13 08:26:05 GMT 2013
                 Start-time Tue Nov  5 13:46:37 GMT 2013
        Total runtime (sec) 671970
      Interval length (sec) 600
            Total intervals 1120
          In-core Intervals 1024

  Summary of the past 1023 intervals
                                     max
            Read Throughput        1.283 MB/s
           Write Throughput        3.239 MB/s
         Cacheable Read (%)      100.000 %
        Cacheable Write (%)       81.064 %
   Max Projected Cache Size      119.937 GB
     Projected Read Offload       96.000 %
    Projected Write Offload       84.000 %

  Summary Cache Hit Rate vs. Cache Size

         Size        20%        40%        60%        80%       100%
     Read Hit     19.000     38.000     57.000     77.000     96.000
    Write Hit     47.000     56.000     64.000     72.000     84.000

  The entire results and output of Automated Workload Analyzer (AWA) are
  estimates. The format, syntax, CLI, results and output of AWA may
  change in future Data ONTAP releases. AWA reports the projected cache
  size in capacity. It does not make recommendations regarding the
  number of data SSDs required. Please follow the guidelines for
  configuring and deploying Flash Pool; that are provided in tools and
  collateral documents. These include verifying the platform cache size
  maximums and minimum number and maximum number of data SSDs.



  ### FP AWA Stats End ###

CAVEATS

This disclaimer is attached at the end of the AWA Summary Output:

The entire results and output of Automated Workload Analyzer (AWA) are estimates. The format, syntax, CLI, results and output of AWA may change in future Data ONTAP releases. AWA reports the projected cache size in capacity. It does not make recommendations regarding the number of data SSDs required. Please follow the guidelines for configuring and deploying Flash Pool; that are provided in tools and collateral documents. These include verifying the platform cache size maximums and minimum number and maximum number of data SSDs.


Table of Contents