Before converting an existing aggregate to a Flash Pool aggregate, you can determine whether the aggregate is I/O bound, and what would be the best Flash Pool cache size for your workload and budget. You can also check whether the cache of an existing Flash Pool aggregate is sized correctly.
In the following example, AWA was run on aggregate "aggr1". Here is the output of the awa print command after AWA had been running for about 3 days (442 10-minute intervals):
### FP AWA Stats ### Basic Information Aggregate aggr1 Current-time Mon Jul 28 16:02:21 CEST 2014 Start-time Thu Jul 31 12:07:07 CEST 2014 Total runtime (sec) 264682 Interval length (sec) 600 Total intervals 442 In-core Intervals 1024 Summary of the past 442 intervals max Read Throughput 39.695 MB/s Write Throughput 17.581 MB/s Cacheable Read (%) 92 % Cacheable Write (%) 83 % Max Projected Cache Size 114 GiB Projected Read Offload 82 % Projected Write Offload 82 % Summary Cache Hit Rate vs. Cache Size Size 20% 40% 60% 80% 100% Read Hit 34 51 66 75 82 Write Hit 35 44 53 62 82 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 ### ________________________________________
The results provide the following pieces of information:
The throughput measurements can help you identify an aggregate that is receiving a higher amount of traffic. Note that these numbers do not indicate whether that aggregate is I/O bound.
The size at which the SSD cache would hold every eligible data block that was requested from disk during the AWA run. Note that this does not guarantee a hit for all future I/O operations, because they might request data that is not in the cache. However, if the workload during the AWA run was a typical one, and if your budget allows for it, this would be an ideal size for your Flash Pool cache.
The approximate percentages of read and write operations that would have been handled by a Flash Pool cache of the optimal size rather than going to disk (projected cache hit rate). Note that this number is related to the performance increase you would see by converting the aggregate to a Flash Pool aggregate, but not an exact prediction.
This table can help you predict the performance impact of decreasing the size of the SSD cache from Max Projected Cache Size. These values are highly impacted by your workload. Depending on whether data that was aged out of the cache was ever accessed again, the impact of decreasing the size of the cache might be large or almost nonexistent. You can use this table to find the "sweet spot" of cost versus performance for your workload and budget.