Datasheet

Solid State Drives for IBM BladeCenter and System x servers 5
Durability
Because flash memory does not have the mechanical limitations of traditional spinning hard drives, SSDs
are less susceptible to shock and vibration and have a higher
tolerance for wide temperature and humidity
ranges.
Low power consumption
Replacing HDDs with SSDs results in a lower cost and greener enterprise. Because there are fewer
storage devices needed, fewer resources (such as controllers, switches, and racks) are needed, resulting
in:
A smaller footprint in the enterprise

Quieter operation

Reduced cooling requirements

Reduced power requirements

Reduced floor space

These reductions result in an overall lower total cost of ownership.
Performance
Because there are no moving parts, startup times are small because no spin-up or seek time is required.
For example, when an HDD retrieves a large file, it searches for the file in passes with each revolution of
the spinning disk, resulting in access times of 10 - 15 ms on average. An SSD can retrieve the same file
as quickly as 0.1 ms. This makes SDD server usage ideal for applications where throughput is more
important than capacity, such as video distribution and financial analysis.
The improved application performance of SSDs results in increased and more reliable transactions in less
time. A comparison of IBM high-performance SSDs with traditional enterprise-level HDDs demonstrates a
dramatic increase in overall I/O operations per second (IOPS), as shown in Table 6.
Table 6. IOPS comparison
HDD (3.5" 15 K) HDD (2.5" 15 K) SLC SSD MLC SSD
Write IOPS 300 250 1600 3000
Read IOPS 390 300 4000 20,000
Cost per IOPS ($) $0.52 (146 GB) $0.83 (146 GB) $0.09 (50 GB) $0.04 (50 GB)
Note: All results with 4 K block transfers
Stated another way, if your application's demands can be met by implementing a large RAID array of
HDDs, you can use far fewer SSDs and achieve the same performance.
Data reliability
SLC and E-MLC solid-state drives utilize several techniques to ensure data stability and retention:
Wear-leveling algorithms that evenly distribute data across the drive.

Garbage collection that uses an algorithm to select the blocks in the memory to erase and rewrite.

For correctable errors, the drives use an ECC scheme (twenty-four 9-bit symbols using Reed

Solomon).