Factors affecting direct attached storage device performance in the application layer

Agents, and HP System Management Homepage when the SSD has reached its maximum rated usage limit

so that you can replace the SSD before it fails.

Duty cycle and I/O workload

Duty cycle and I/O workload affect drive reliability. This section briefly discusses how to calculate duty

cycle and I/O workload, and the influence they have on drive reliability.

Duty cycle (DC) is the number of hours the storage device is powered-on, divided by the number of total

hours in a calendar year (8760 hours).

For example, HDDs powered-on and spinning for 8,000 hours during a calendar year may see more

failures due to overheating than HDDs powered-on and spinning for 870 hours. The calculated DC for each

operating period is as follows:

 DC

(8,000)

: powered-on for 8,000 hours in a calendar year would have a DC of 91% or

8,000 hr/8,760 hr.

 DC

(870)

: powered-on for 870 hours in a calendar year would have a DC of 10% or 870 hr/8,760 hr.

I/O workload (WL) is the number of hours the storage device actively reads or writes data, divided by the

number of total hours in a calendar year.

For example, as HDD read/write workload increases from 2,300 hours to 6,000 hours, the stress and

failure of mechanical parts (such as spindle, drive head, or motor) may increase. The calculated WL for

each operating period is as follows:

 WL

(2,300)

: active read/write for 2,300 hours in a calendar year would equal a WL of 26% or

2,300 hr/8,760 hr.

 WL

6000)

: active read/write for 6,000 hours in a calendar year would equal a WL of 68% or

6,000 hr/8,760 hr.

Calculating DC and WL for SSDs is the same as for HDDs, but wear-out is a more meaningful metric for

SSDs, as previously described in the Wear protection technology section.

Application environment

Storage needs of applications vary. As indicated in Figure 1, each application has unique requirements for

throughput and I/O workload. Table 3 identifies some types of applications that place heavy demands

upon storage. For example, oil and gas companies rely upon detailed seismic analysis to estimate yield

and income from a future gas field. Web based companies require the ability to quickly track, store, and

mine user behavior for targeted advertising and content. Financial service corporations need to analyze

multiple sources of structured market data combined with unstructured new accounts quickly to price

investments properly in near time or real time. These industries have key challenges in common: balancing

the speed of computation, the speed of storage processing, and the total storage costs that yield profitable

results.