There are many metrics that can be used to measure the performance of a storage system. Amongst all, two metrics specifically give the most insight into the performance of the storage system:
The operations a system performs are a direct function of the client operations requested by applications within an enterprise. The operations requested by an application are referred to as an 'application workload', often shortened to simply 'workload'. Workload characteristics that can affect and describe performance include the following:
- Operation type – [Reads/Writes/Other]
- Randomness/Sequence of workload
- Concurrency (parallelism) of workload
- Scheduled background tasks/jobs
Changes in any of these workload characteristics can affect the potential performance of the system, and can be observed in either latency or throughput. Over time, workload almost always increases organically. It is important to note this when reviewing current-day system performance.
For the purposes of day-to-day management, there are a few guiding principles behind performance. These can be stated as the relationships between the fundamental characteristics of a workload and the resulting performance:
- Throughput is a function of latency
- Throughput is a function of concurrency
- Throughput is a function of operation size
- Throughput is a function of randomness of operations
- The host application controls the amount and type of operations
Every storage system has a performance limit, or ‘budget’. To maintain good performance, it is important to stay within the guidelines of that budget. Every feature and option has a performance cost, and it is important to note this when configuring and managing your storage system.