Learning About SSDs

HP-Pavilion-HPE-H8-Desktop-PC-300px.pngSolid state disks (SSD) are coming down in price and going up in capacity. The attractions are many; lower power consumption, low heat output, mechanically robust, decent write performance and dramatically faster read performance. There’s plainly a lot to like about SSDs.

Last winter I put a cheap 120 GB San Disk Ultra SSD into my aging netbook and gave it another year’s lease on life. Over the summer I saw a deal on some nicely spec’d HP Pavilion HPE desktops I bought a couple for myself and the Mrs. It seemed a sensible way to move us away from Windows XP.

This is a little story about the solid state disk residing in my desktop PC. The device in question is a 128 GB Crucial M4 model that I added to a new HP desktop purchased from Woot.com last summer. The tale is worth telling because the SSD seemed to fail after just a few months.

Of course, the HP didn’t come with an SSD. It came with a 2 TB Seagate ST2000, a fairly standard, if largish hard drive. I decided that I would add the SSD,  making it the boot volume.

Some might ask why do this? It’s a valid question. In my case the answer is two-fold; primarily to satisfy my curiosity, but also to gain some greater experience using SSDs. My employer is beginning to ship systems that boot from SSD-based storage, so further experience with their use, even their failure, could be valuable.

Cloning the boot volume to the SSD was easy enough. You don’t even need much drive space to hold just the OS and application data. I then setup Windows to use the 2TB drive for the user profile data. I found a simple tutorial on how to create exactly this arrangement.

The upgraded system was a joy to use. It boots the OS very quickly, yet it still has all the storage that I need for my various work and personal activities.

Crucial-M4-SSD.jpg

Four months after this little exercise I rebooted the system in the process of installing an OS update…but it did not reboot. In fact, it did not even sense the presence of the SSD at the BIOS level. I thought to myself, this is very bad. Very bad indeed.

Fearing the worst, I removed the SSD from the HP desktop and connected it to my laptop using an external SATA-to-USB adapter. The laptop did not acknowledge the presence of the SSD. Nor did the adapter show any drive presence or activity. Not good at all.

Thinking I was at the end of my diagnostic capabilities I visited the manufacturer’s web site. There I found a chat-based support tool. I engaged one of their staff, informing him of my situation. He asked a few questions and shortly determined that I would need to return the drive under warranty.

Given that advice I found a 160 GB WD Scorpio Black hard drive that I had earlier removed from my netbook. I was able to load it with a backup image of the new desktop, restoring the desktop to an operational state, although without the SSD.

Next I called the Crucial Technologies support line thinking that I would seek an RMA number to return the SSD under warranty. I’m very glad that I made that call.

The technician who took the call asked me all the questions I expected, but did not reach the same conclusion as the prior technician. He informed me that they would not accept the drive under warranty until I had allowed it to run through its “maintenance mode.”

Apparently, when the drive is powered but not connected to a SATA controller, it will drop into maintenance mode after 8-10 hours. Once in this mode the drive controller goes through an internal process of checking itself against the storage media looking for inconsistencies.

He also told me that there was a firmware update for the device. Of course, to install the update I first had to get it into a state where it would be recognized by a host. Since I already had the drive in an external mount I could simply left it powered overnight and hoped for the best.

The next morning I once again connected the SSD to my laptop using the SATA-to-USB adapter. Initially the laptop did not recognize the newly attached device, which I took as a bad sign.

I was distracted for a few minutes by a phone call. As I was on that call out of the corner of my eye I saw the laptop display the familiar prompt for display of newly attached media. The SSD has returned from its long sleep.

Subsequent to that I was able to update the drive firmware. I’ve since re-installed it to my desktop and it has been operating normally for the past few weeks.

The experience causes me to ponder the reality of using SSDs. They are clearly more complex than the magnetic media they replace. Their failure modes are very different. Sometimes a drive that appears to have failed, as was the case with mine, may simply be suffering an unfortunate problem in firmware.

  • Straywasp

    This is interesting. I’ve not heard of that problem before. If modern hard drives are anything to go by then the firmware can be extremely complex.

    • mjgraves

      Being involved in broadcast production I can remember when non-linear editors first came on-scene. I remember disk firmware being an issue with respect to thermal re-calibration. Some manufacturers “zoned out” the inner drive sectors to ensure more constant drive throughput at the cost of drive capacity.

      More recently I’ve had years of using flash media for embedded systems. I’m familiar with such issues as wear-leveling. I expect that the controllers for SSDs are as complicated as those for traditional disks, but in different ways.