|Overall Statistics||WinRAR: Compressing|
|Data Read||1.51 MB|
|Data Written||5.60 MB|
|Disk Busy Time||0.03 s|
|Average Data Rate||213.55 MB/s|
When you compress data, you're essentially taking multiple files and assembling them into a single file that occupies less space. As a result, you might assume that this task would consist of sequential accesses. The opposite actually turns out to be true.
Why is this? All forms of data compression replace repeated occurrences of data with references to a single copy of that data existing earlier in the input (uncompressed) data stream.
Consequently, compression turns out to be mostly a random write-heavy operation. However, the amount of data the system reads and writes is greater than the sum of the uncompressed (~400 KB) and compressed files (1 MB).
- 58% of all operations occur at a queue depth of one
- 37% of all operations occur between a queue depth of two and eight
- 11% of all data transferred is sequential
- 12% of all operations are sequential
- 29% of all operations are 4 KB in transfer size
- 32% of all operations are 32 KB in transfer size
- 14% of all operations are 128 KB in transfer size
- 13% of all operations are 16 KB in transfer size
I suppose I can see some inexpensive reliable SSDs in office machines in the near future, mostly to reduce the failures connected with mechanical drives and speed up boot times and installation times.
SSD not really appropriate.
Unless by office computer you mean where you have the only computer in the office, or files do not need to be shared around the office.
Given the amount of work people do who open large files (where an ssd may be appropriate), they are too small/too expensive to be justified.
Example, large 3d CaD drawings, spend extra money on them loading faster, lose funds for better overall computer (graphics especially).
I find it ironic that the only place your tiny ssd drives are good enough are in computers where speed isn't important in the first place. Until 320GB ssd's can compete with regular magnetic drives, it isn't an option to upgrade.
Imagine the added cost of upgrading the 2442 registered clients to ssd drives! About half could make do with a 120GB drive, and the rest would need at least 160GB and possibly bigger.
That's an expense you can't possibly gain in productivity.
Replacing sas drives with ssd's might make sense for your database or vmware/hyper-v systems, but it isn't going to make much sense on the majority of workstations.