Performance

The performance chart (heat map) provides information about the:

  • Periods of time when the performance is identified as outliers (latency is unusually high) for a particular workload.

  • Periods of time when the hosts must have experienced high storage latency.

  • Performance insights.

Latency: Latency (response time or the time taken to service an IO) is a measure of the performance of the system and depends on the:

  • Storage system configuration

  • Workload characteristics

  • Current load

If the parameters that influence the metrics are known, you can calculate an approximate latency.

Performance/Latency score: Latency score is a comparison between the following:

  • Measured latency of a storage system at a given point in time.

  • Latency observed across storage systems with similar configuration and workload in the install base or the field.

    The comparison determines if the value of the metric (latency) is optimal or unusual (an outlier). The latency score measures the relative severity of the latency that is reported by the storage system in comparison to the latency observed in the field. The performance chart reports the latency score for observations at 5 minute intervals. The observations are reported as column charts.

    The latency score ranges from 0 to 10, where:

    • 0 indicates that the latency reported by the storage system is observed in the field.

    • 10 indicates that the latency is unusually high for the given workload characteristics.

Exceptions for the performance score (latency score)

  • The performance score for a storage system reaching 100% saturation must be higher than 0 for the observations at that time. However, this score is not mandatory and a performance score 0 could indicate that the cache hit is beyond 50%.

  • If the response time is expected to be in the similar range as observed in the field, the score might be still reported as 0 or a value closer to 0.

  • If the storage system has a configuration which is resource-intensive (too many compressed volumes), the latency recorded might still be an outlier compared to the typical latency for the specific type of workload. In such scenarios, the performance score can be a higher value which is closer to 10.

For more information on performance insights, see HPE SSMC User Guide.