6.4. Raid storage performance and reliability factors
- Application I/O work-load pattern
- Bandwidth
- Cost
- Capacity
- Performance
- Scalability
- Reliability
- Power consumption
IOPS-intensive applications
require the storage system to process as many hosts read and write requests (or I/O requests) per second as possible, given the average I/O request size used by this application, which is typically 4 - 8 KB. This behavior is most common for OLTP databases.
Bandwidth-intensive applications
require the storage system to transfer as many gigabytes of information per second as possible to or from hosts. They typically use an I/O request size of 128 - 512 KB, or more. These characteristics are commonly inherent to file servers, media streaming, and backup.
- input/output requests per second (IOPS)
- throughput (measured in MBps or GBps), depending on application workload
System crashes
Events such as power outages, os crashes or hardware failure. This can affect write operations causing partial writes or parity inconsistencies.
Bit errors
Bit errors result from incorrect data writes. Bit errors are only discovered when the data is attempted to be read.
Multiple disc failures
Events such as power failures/surges, temperature or physical stress can lead to a correlated failure of multiple discs. Discs are most probable to fail either at the beginning or at the end of their lifetimes.