SSD, decent boot drive for $99.00

no way am i dealing with a storage drive that is known to fail. Ive got enough gray hairs and bald spots from the ones that are normal. WTF. Thats like buying a car that is prone to blow up.

LOL!
All drives will fail in time.
I suggested the SSD for boot and a second HDD for storage until the bugs are worked out of the SSD.
 
It sounds great.

I'm wondering about the swap file. SSD's have a limited write capability. The computer is frequently writing to the swap file on the drive. I wonder if that will shorten the life of these drives or have they addressed this problem somehow.

Either shut off the swap file, or move it to a different drive. With 2GB (XP) and 4GB (Vista/7) of RAM, your system should not need the swap file. Essentially, if the system swaps, it probably means some program is trying to consume all available memory, usually causing either the program or the computer to crash. The swap file is only providing more memory to extend the time before this death takes place.

It used to be that some people actually used the swap file for virtual memory, but lately memory is so cheap everyone seems to have enough of it. Some users really do need virtual memory, but they tend to be power users that already know that are familiar with the special situation they are in.
 
All HDD/SSDs will fail at some point. Yes SSDs have a set limit to them, but are pretty stable, and old school HDDs have moving parts that can break. The post mentioning the SSDs failed was using them in a heavily used database server, which would have way more reads/writes than any home PC would use. Plus check out this article, SSDs are getting faster, and longer lifespans. http://www.anandtech.com/storage/showdoc.aspx?i=3688 Give the SSD two more years and things will be looking up. I am a little sad that flash had to pop it's head into the SSD market though, magnetic ram is sooooo much better, but too expensive.
 
SSD were touted to have a 100 year life when first introduced!
SSD Manufacturers said they would easily outlast PCs!
Guess they were wrong....:o

Who said that? Certainly nothing I ever saw suggested the devices have a long life unless written only rarely.

For some reason, 10k write cycle lifespan comes to mind for SSD devices, much like USB thumbdrives.
 
Who said that? Certainly nothing I ever saw suggested the devices have a long life unless written only rarely.

When they first came out there were a few articles saying they would last that long. Here is one of them reports.

Solid-State Drives, SSD, are coming out this spring to replace the existing HDD. Dell is offering them in some notebooks now, while HP will follow this spring. What seems most impressive are claims the drive will offer a mean time between failures (MTBF) of one million hours.....that`s about 114 years!

Samsung, Toshiba, HP and others are working on them
Here`s the Samsung 128GB SSD version:

Samsung spins 128GB SSD

https://terb.cc/vbulletin/showthread.php?t=180780
 
we have 2.5 s sitting in 3.5 drive bays they are just not bolted in ... main reason for the physical mounting is spin... no spin on the SSDs.
We were using them for MySQL journals and they were super fast .. unitl we found out about their limited life span. We were using the IBM M Series Extremes for about 1600 bucks..... and after 5 monthes we have had two with catastrophic failures ...... now i popped one in my PS3 and it made little difference, but the one i have in my laptop.... wow it boots instantly.
One thing we did find was not all controllers recognise the drives. OCZ drives seems to work across the board but i found for the cost only about a 1 to 3 % speed boost over a good SAS drive. Again this is for DB work.
But as far as boot in my laptop it rocks.. but do take heed they do have a limitted life of i think something like 10000 writes or so.

I do this at work too ..... sorry not IBM drives Intels ...
 
When they first came out there were a few articles saying they would last that long. Here is one of them reports.

Thanks for the links.

I can see how someone could confuse a theoretical MTBF, or even the theoretical write-cycle lifespan of a drive, with its actual useful lifespan.

Even if you search for articles on SSD write cycle life span, you'll find a number of folks who think that nobody actually stores any data on a hard drive (or an SSD): they all work from the assumption that writes are evenly smoothed across the SSD chips, completely ignoring the fact that the majority of an SSD will contain static data - and that write cycles will be focused on a much smaller "free space" on the drive.

And because of the enhanced wear algorithms which have been developed, the time difference between first block failure and catastrophic failure of the drive will be very small: even though 90% of the drive may be perfectly fine (as the content has been quite static), the 10% which is used extensively will likely fail completely in short order.
 
And because of the enhanced wear algorithms which have been developed, the time difference between first block failure and catastrophic failure of the drive will be very small: even though 90% of the drive may be perfectly fine (as the content has been quite static), the 10% which is used extensively will likely fail completely in short order.


I may not understand SSD's all that well; still, I was under the impression that the controllers on SSD's are designed to ensure that there is no "10% of the drive that is used extensively".

Ie - the controllers on SSD's constantly ensure that use of the drive is broadly dispersed evenly across all sectors of the drive. Ie the controllers on SSD's employ "wear levelling" technology to ensure that all sectors of drive are equally utilized. Also if the controller determines that a certain part of the drive looks 'unhealthy' it flat out never writes to that sector to enhance product lifespan.

USB flash drives generally don't have controllers that ensure "wear levelling"; so, these flash memory devices can have sectors that are heavily used and hence lead to the unit burning out earlier.

http://www.storagesearch.com/siliconsys-art1.html

This is the main reason why SSD engineers recommend that you never defragment an SSD drive, - because defragmenting kind of sabotages the controllers efforts to ensure that use of the drive is directed at healthiest sectors of the drive.
 
I may not understand SSD's all that well; still, I was under the impression that the controllers on SSD's are designed to ensure that there is no "10% of the drive that is used extensively".

No, the controllers on SSDs are designed to ensure that new writes are directed to blocks with lower write-cycle counts.

They do nothing to the files you've already got on the device, some of which will be in blocks with write-cycle counts in the tens or less.

Let me try it this way: you format a new SSD, and install an OS on it. Those OS files might never get changed, and will always sit there with a very low write-cycle count.

Then you start to fill the drive with a few pictures, files, etc. Those files may come and go, and will be in blocks which move into low-write-count -but available- sections of the drive as they're written.

The drive fills to, say, 90% full. Now, the temporary and short-lived files which were always getting written are constrained to that remaining 10% of free space - in effect, shortening the lifespan of the overall device by a factor of 10, in this example.

If you delete a few pictures, that space will be freed up again and will be used preferentially for new files (since it has a low write count). But it could be another picture that fills the space, or it could be temporary files. And there's no mechanism for selecting the blocks which have the shortest remaining lifecycle for a file which might survive a longer while. Instead, the mechanisms will tend to put the next picture back into the available blocks with the lower write count. So as the drive goes back to 90% it'll be mostly the blocks with the higher write count which are still being used for those temporary files, killing the device efficiently.
 
Oh, that's a good point.
I guess the moral is make sure your SSD doesn't go over the 85% capacity mark or so, so that the controller doesn't get painted into a corner where it has to use assets it would rather not use.
 
Anynym points out some great points.
It looks like this 30GB SSD is ideal for both Linux and OSX because they require no defragging. I planned using it for a boot drive so Linux would only require < 5GB leaving the balance for temp files. Any files needed to be saved would go directly to the 1 TB HDD so as to not unduly shorten the life cycle of the SSD.
 
I'd expect that we'll see some hybrid drives developed with both solid state and DRAM devices, and file systems developed to explicity recognize transient file systems. (E.g. a differentiator for the Unix "/tmp" file system).

This would allow systems to put transient files into DRAM and longer-lived files into SSD. A battery back-up could even allow some files to be transferred from the DRAM into the SSD on power failure (prioritized by an algorithm within the Temp Filesystem).

But we aren't there yet either.
 
I'd expect that we'll see some hybrid drives developed with both solid state and DRAM devices, and file systems developed to explicity recognize transient file systems. (E.g. a differentiator for the Unix "/tmp" file system).

This would allow systems to put transient files into DRAM and longer-lived files into SSD. A battery back-up could even allow some files to be transferred from the DRAM into the SSD on power failure (prioritized by an algorithm within the Temp Filesystem).

But we aren't there yet either.

Do us a favor and start sending detailed emails to all the drive makers please!

There are SSDs out there that are made with magnetic DRAM the problem is the last time i seen a price listed for it it was $2200 for a 6gb drive.