MOTUnation.com

HCMarkus wrote:

RecordingArts wrote:I'm a big fan of the 2.5" 10,000 RPM 300Gig Western Digital VelociRaptors but just had 2 fail in a matter of 2 weeks, and then the 2 replacements failed too! Not sure what is going on here. WD is sending more replacements but it gets a little scary when stuff like this starts happening.

Check your HD power supply... especially if the Raptors are in external cases! When supplies get old, they sometimes don't put out rated voltage/amperage, and drives can't spin up properly. I posted awhile back about how I replaced a failing external drive power supply with a stock PC power supply, $40 at Fry's. It is now powering four externals, (including two Raptors) with plenty of headroom for more.

HCMarkus wrote:SSD!

RecordingArts wrote:I'm a big fan of the 2.5" 10,000 RPM 300Gig Western Digital VelociRaptors but just had 2 fail in a matter of 2 weeks, and then the 2 replacements failed too! Not sure what is going on here. WD is sending more replacements but it gets a little scary when stuff like this starts happening.

newrigel wrote:SSD! Man... if you have the extra money... definitely YES!

Depending on the type of NAND flash memory (SLC or MLC), and the predicted usage per day, one can estimate the overall lifetime of the SSD (sometimes called the write endurance of the SSD). However, it is particularly important to emphasize usage per day. Embedded in the word usage and often overlooked is the type of usage, such as sequentially written, random written or mixed written (random and sequential). Each scenario affects performance (IOPS, transfer rate, block size, etc.) and the overhead written along with the data. These parameters, when accounted for properly, can effect the resulting life time estimation, even when the total perceived usage per day is the same. This is why, when life time estimation is quoted, the vendor indicates the type of written strategy executed, which is usually sequential write. Sequentially written data provides an ideal scenario for estimating lifetime since the IOPS, transfer rate and overhead are constant factors that can be used in a simple formula, as opposed to random or mixed written data, which can vary the overhead size. The calculated lifetime estimation example used in this paper will exercise the sequential written usage type, which is a reasonable estimation for the SSD lifetime, and will give the reader a base of understanding for the predicted value. Keep in mind, the true written usage will vary depending on the type of write strategy used.
Recall for the SLC NAND flash, which is the present technology used today, P/E cycle endurance is approximately 100,000 cycles, as opposed to MLC which is only 10,000 cycles.
In all cases the equation for lifetime is based on the simple product of SSD capacity and the write endurance of the NAND flash memory. This is because the wear-leveling algorithm insures the wear rate for each block of flash
memory to be essentially equal throughout the entire lifetime of the drive. The plot below shows the wear leveling result after constantly writing to a 32GB Imation SSD for 170 hours. Note each block has about the same number of erase counts.

ripeart wrote:

newrigel wrote:SSD! Man... if you have the extra money... definitely YES!

I second this, however keep in mind that SSD drives have a limited number of reads/writes before they fail. Here's a quote from a whitesheet Imation published (late 08).

http://www.imation.com/PageFiles/83/SSD ... -Paper.pdf

Depending on the type of NAND flash memory (SLC or MLC), and the predicted usage per day, one can estimate the overall lifetime of the SSD (sometimes called the write endurance of the SSD). However, it is particularly important to emphasize usage per day. Embedded in the word usage and often overlooked is the type of usage, such as sequentially written, random written or mixed written (random and sequential). Each scenario affects performance (IOPS, transfer rate, block size, etc.) and the overhead written along with the data. These parameters, when accounted for properly, can effect the resulting life time estimation, even when the total perceived usage per day is the same. This is why, when life time estimation is quoted, the vendor indicates the type of written strategy executed, which is usually sequential write. Sequentially written data provides an ideal scenario for estimating lifetime since the IOPS, transfer rate and overhead are constant factors that can be used in a simple formula, as opposed to random or mixed written data, which can vary the overhead size. The calculated lifetime estimation example used in this paper will exercise the sequential written usage type, which is a reasonable estimation for the SSD lifetime, and will give the reader a base of understanding for the predicted value. Keep in mind, the true written usage will vary depending on the type of write strategy used.
Recall for the SLC NAND flash, which is the present technology used today, P/E cycle endurance is approximately 100,000 cycles, as opposed to MLC which is only 10,000 cycles.
In all cases the equation for lifetime is based on the simple product of SSD capacity and the write endurance of the NAND flash memory. This is because the wear-leveling algorithm insures the wear rate for each block of flash
memory to be essentially equal throughout the entire lifetime of the drive. The plot below shows the wear leveling result after constantly writing to a 32GB Imation SSD for 170 hours. Note each block has about the same number of erase counts.

newrigel wrote: Plus, with 320 GB in read writes daily equals lasting 27 years... that's just fine with me!
5 years from a drive is quite a bit!

MIDI Life Crisis wrote:Seagates can be a bit shaky, as some WD other colors.