Solid State Drive story.

Are consumer solid state drives ready to be used in production environments? The newest generations possibly are. During the last years, however, in my opinion they clearly were not. Let me share my SSD horror story, beginning in 2009.

Disk 1
It was the end of 2009 when the SSD boom reached me and I decided to replace the classical HDD in my notebook with a 128 GB Super Talent UltraDrive GX (FTM28GX25H). This drive was a recommendation in many hardware forums. In tests (e.g. at AnandTech, Tom’s Hardware) it showed a very good performance. It is Indilinx Barefoot controller-based and built with MLC memory — as almost any other consumer SSD at these times. This was my SSD 1, with 5 years of warranty and 10 years of data integrity (ha).

Disk 2
I was just having performance fun with my new SSD for a few days, when I realized that the particular drive series I had bought had a performance issue. People said Super Talent selected cheaper flash chips in its XXXX-series. Whatever they did, the I/O performance dropped significantly. I wrote a blog post about this back in 2009. I registered with a Super Talent reseller and managed to receive a replacement product — free of charge, of course. The same model, a different revision. This was my SSD 2.

Disk 3
I had a few weeks of fun with SSD 2, did not put it under heavy load, almost did not write to that disk. One day, the BIOS of my notebook just did not recognize the SATA device anymore. The disk was gone. Contacting the shop brought me another replacement: the Extrememory XLR8 Plus, one of the disks with the (at these times) new, faster and more reliable Sandforce SF-1200 controller (obviously I did not want to have a third version of the Super Talent UltraDrive GX). This XLR8 Plus was my SSD number 3.

Disk 4
The XLR8 went fine for some time. I don’t remember exactly. Let it be 2 months, must have been spring 2010. What happened then was just the same as during the last iteration. Suddenly, the SATA device was not recognizable anymore. The disk was gone. I contacted the hardware shop again and got a replacement disk — free of charge, of course. This was disk number 4. An XLR8, again. Great, awesome, plug it in, use it and have some more high performance fun for a random number of days! No.

Every time when an SSD was busted, I immediately went back to the classical magnetic HDD until the replacement arrived. At this point, if you like, imagine the pain of fiddling around with the notebook every time I had to change the disk. Think of those kinds of screws you can only unscrew and screw back in three or four times until the plastic thread is done. Imagine FreeDOS-based SSD firmware updates, frequent operating system setups, data migration scenarios. Maybe you can understand that this time I just left the classical HDD in the machine. I did not even unpack the brand-new XLR8. I kept it in the drawer for a year.

Disk 5
Spring 2011. Same notebook as in 2009. It had an optical disk drive. At some point I bought one of these cool HDD caddies. I removed the optical drive and installed the 1-year-old brand new XLR8 in my notebook — as a second disk. I used it for some time until… this time it was ongoing, creepy data corruption. SSD 4 was busted, too. I contacted the vendor, Extrememory, and got a replacement — free of charge, once again. After making clear that I do not want to have the same model again, they organized for me to receive a 120 GB Kingston V+200 which is sold in both, consumer and business categories. Awesome and uncomplicated service. This is my SSD number 5. It is based on the Sandforce SF-2281 controller, three generations newer than the Indilinx Barefoot that was in the SSD which I initially paid for.

Current state
For half a year I am running my production system (based on the same old notebook) with two disks. The V+200 enormously speeds up the operating system and applications, while the classical HDD is a good old data container. No problems with this setup so far.


  • Whatever they advertised years ago, when the SSD consumer products came up, about data integrity and mean time between failures: all this likely was bullshit. These numbers as well as the warranty durations were kept high in order to gain customers — like me. I think all of these companies were well-aware of the fact that their products are still experimental and might fail for various reasons. Still, they all were fighting for their market share.
  • The hardware might have been tremendously bad. The service, however, was well-functioning. I paid for one drive and got 5, whereas the last one is three generations newer than the initial one. That’s what I would expect of an industry dealing with unstable hardware.
  • Maybe now the time has finally come where you can use a consumer solid state drive with the same confidence in your data and in the reliability of your production system as with classical HDDs.
  • It’s always the same and it really does not matter how crappy your disk device is: you have to properly back up your data.