Description
|
Calculates the system's I/O workload, and shows the relationship between the workload and the amount of storage capacity used.
This data reveals how much capacity is frequently accessed over time ("hot"). You can use this information to determine how system performance may benefit from implementing a tier of fast SSDs, instead of slower spinning disks, for some or all of that capacity. Users often see the greatest performance benefits when the SSD tier is sized to handle 80% or more of the I/O workload.
Calculations are based on user-specified settings and up to eight days of usage data captured by the system. The storage system must be in a stable state for a minimum of two days to generate complete data.
For a graphical view of workload data, see the I/O workload panel in the PowerVault Manager.
The workload history does not persist beyond a controller restart.
NOTE:The suggested capacities may not apply to heavily streaming workloads.
|
Parameters
|
calc-type peak|average
Specifies whether to base the calculations on either the peak values saved in the usage data or the average values. For calculations, the pool is divided into equal bins of LBAs. Each sample contains readings for all bins. There are multiple samples taken per day. To calculate average, the sum of the readings of a bin are divided by the number of samples. To calculate peak, the largest bin value from the collection of samples is taken, instead. This leaves one value for each bin whether average or peak was selected. From there, workload calculations are made using the bin numbers as input.
io-type reads|writes|combined
Specifies to limit the data used for calculations to small read I/Os only, small write I/Os only, or the combined total of small read and small write I/Os. Small I/Os are random access operations, as opposed to large I/Os which are sequential access operations.
controller a|b
Specifies whether to base calculations on data from the pool owned by controller A or B.
target-pct
target-percentages
Optional. Default workload calculations are based on low, mid, and high percentages of capacity: 50%, 80%, and 100%. This parameter overrides any or all calculations with your own percentages expressed as whole numbers. Enter a comma-separated list of up to three values. If a value is not specified, the corresponding default will be used. For example, entering 65,85,95 would set percentages to 65%, 80%, and 95%. Entering % characters is optional.
|
Output
|
Current SSD Space
Current SSD capacity allocated to the pool formatted to use the current base, precision, and units.
Pool
The pool for which the calculations are based: A or B.
Calculation Type
Either Peak or Average.
I/O Type
Calculations are based on either Reads, Writes, or the Combined total of reads and writes.
For each data sample:
Start Sample Time: Datestamp for the first data sample used in calculations.
End Sample Time: Datestamp for the last data sample used in calculations.
<low%>, <mid%>, <high%>: Columns showing calculated capacities based on the three percentage values specified in the
target-pct parameter or their corresponding defaults. The column headings show the respective percentage values.
If less than 90% of the expected data samples are available for calculation, the calculated capacity will be
N/A. Samples may be unavailable for rare reasons including controller failover conditions.
|
Examples
|
Calculate the peak workload of small read I/Os for the pool owned by controller A, with a low capacity target of 70% instead of the default 50%. The results indicate that over the time and capacity ranges, about 1600GB of storage is frequently accessed, making it a good candidate for use of SSDs.
# show workload calc-type peak controller a io-type reads target-pct 70
|