Flash Accel Installation and Administration Guide
:
Viewing Flash Accel performance statistics
: Flash Accel metrics
Flash Accel metrics
Flash Accel
collects metrics so that you can view system performance. You must enable performance statistics on each VM and disk that you want to monitor.
The following metrics are collected:
Metric
Description
Cache latency
The average read latency (in milliseconds) per incoming read operation when the item was found in the cache.
Hit rate %
The percentage of read requests that were serviced from cache instead of having to access the storage system.
Insertion rate %
The percentage of read requests that were not accessed in cache and had to be serviced from the storage system. These items are then inserted into the cache.
Population %
The percentage of the cache that is currently being used (based on the total amount that was allocated to the virtual machine). If this number is very low then you may have over-allocated cache to the VM.
Read latency
The average read latency (in milliseconds) per incoming read operation.
Read ops
The total number of read operations.
Write latency
The average write latency (in milliseconds) per incoming write operation.
Write ops
The total number of write operations.
The hit rate number shows the percentage of requests that were returned from the cache instead of having to access the information on the storage system over the network. After the cache has "warmed," the cache hit rate should be relatively high and level off to a steady state. If the cache hit rate drops significantly, there may be an issue that needs to be addressed.
The insertion rate typically starts out very high as the cache is populated. After the cache has been fully warmed (when the cache has been used for awhile), the insertion rate should then be fairly low because most read requests are serviced from cache (appearing in the hit rate at that point).
Parent Topic:
Viewing Flash Accel performance statistics
Was this topic helpful? Send
comments
to NetApp.