Gee, SSDs will be more popular than rotating drives when the price/capacity difference levels out in [insert spurious year here] because they use less power and are faster.

Wow, who'd have thunk it? No, really, I *never* saw that one coming.

In other shocking news, when water gives the same effects as beer with no side-effects and the same low, low prices, beer sales will drop off.

In other news, the sun comes up every morning, and British Summers are ridiculed around the work for being the sweaty tuft of rectal pubic hair they are....

Steven R

When pigs fly?

Won't work on my flash drive. Yes, there are lots of wear leveling goodies, but currently magnetic media works quite well and doesn't have a built-in wear mechanism. Then there are the database files that are updated on a second by second basis that will waste a flash drive in no time.

Wonderful for use on the EEEPC, but for main storage, (I could be wrong now, but) I don't think so. (Its a jungle out there).

p.s. Vista probably needs a big swap file anyway!

that's the "mixed environments", spinney things for the vista desktops to put their swap file on, and flash drives for the servers to save files on (which of course all run *nix with sufficient RAM so should pretty much never swap during normal operation)

mine's the one with the spare RAM sticking out of the pockets

Anything to reduce the amount of noise EMC boxes make is a good move. Data centres are loud enough as it is.

That might be true of your average USB pen drive, but it's not true of SSD. If it were, then what would be the point of GB's of RAM? That gets written to billions of times and I've yet to see a memory chip fail cause it was 'tired'. That's not to say memory doesn't fail, but your mistaken belief would have the stuff falling over in heaps, which patently isn't the case.

with RAM as cheap as chips, why swap at all?

Flippant I know.

If you can, get yourself a copy of EMC's presentation on how they think flash drives, using single layer and all sorts of other whiz tech will actually outlast typical brown, round, spinning stuff.

The estimated lifespan for flash drives is already into durations after which you'll be throwing it away and upgrading anyway, unless you tend to keep hard drives for 10 years that is.

With DDR memory cost a tenner per 1GB (we are talking about 2010), why do you want swap at all? Also, there is no say that SSD will replace HDD entirely. But, given the choice, I would prefer former one even if it were to wear out in 5 years. And HDDs do wear, too.

Flash and RAM chips use different storage techniques; RAM will last a lot longer as it's designed to run at ridiculous speeds and be non-volatile.

And SSDs and USB-sticks aren't the same either. An SSD is far, far more expensive but lasts far, far longer. I believe the eeePC's SSD will last for a good few years of constant reading and writing before it gets noticeably worse.

A REAL computer doesnt need swap space.

I run heavy duty stuff on my workstation, including AutoCAD (memory intensive with big DWGs, in fact the first version I used ran on DOS and had it's OWN swap file) and several 3D packages on Windows XP. With 3GB RAM and no swap file, it hasn't yet run out of memory.

Servers I always spec so they don't need a swap file/partition.

With the cost of RAM now (about £20 for a 2GB DDR2-666 DIMM) adding 8GB to a low-spec (desktop-component-based) server is a no brainer.

What's the reliability figures for flash storage? I've seen far more flash drives fail this year than hard drives.

Flash memory is so small, and with speeds and size always on the up and up.

People are often wiring in an internal SD reader and card to boot up the OS quickly.

Asus have gone one stage further, and placed flash on all their motherboards to load Splashtop in about 5 seconds. Splashtop is a cut down Linux OS that does email and browsing. Linux adoption is set to go through the roof. I cannot wait to see MS try and claim a distinction - how many unix folks have been counted just because it was nigh impossible to get a laptop without getting windows as well.

Swap is not a problem on a SSD if you want to dedicate the space. Swap is hardly ever used nowadays, really it is a safety net, so not too much concern over the finite erase write cycles. If you are swapping you have a problem, developers do it all the time, when they are coding, we call it a bug in the main (infinite loops or memory leaks). But of course some operations do use huge amounts of memory; rendering can do it, as well some more complex mathematical operations where gobs of data need to be accessible.

I would certainly give some space to swap if size is not an issue. On a sub 4GB device perhaps not, but at about 16GB I would. And of course it is trivial to get swap on if you think you need it.

The MBR and a limited area for logging is the problem, as is malware dedicated to blowing sections like the MBR on flash devices.

There is not much of difference between SSD and a USB flash drive memory, as far as the erase write cycles go, in my opinion. And the speeds can be slower on a SSD as well.

There is an issue with wear leveling, and replacement with these devices, but in normal operation it will take quite a few years to see the problem. Really you want to be able to replace the devices with ease, that's the ultimate solution and the one I suspect the market would prefer - so, how many admins does it take to replace a bit of flash memory?

So I assume none of you have ever seen a 50Tb DB2 database then? I have boxes with hundreds of Gb of RAM. You still need to swap if the app is big enough.

Users of EMC storage don't generally hook it up to a PC with a 'massive' 8Gb of RAM. They hook it up to E25Ks, P690 Regattas and other such beasts.

Oh, Dr Mouse - if it's running Windoof it isn't a server. No, really, it's not.

One of the real benefits of enterprise flash is that the working life is completely predictable, unlike the brown spinny stuff. This means that many RAID techniques will be redundant and you can plan for maintenance as you know exactly when you will need to replace your SSD drives

http://www.storagesearch.com/ssdmyths-endurance.html

A lot comes down to the distribution algorithm and the files you're writing.

From a USB drive (Windows) user point of view - I hate that FAT is the default (4GB file size limit is no good for Virtual PC files) and that you can't partition them (I need a 4GB FAT partition and the rest as NTFS).

The numbers and density for SSD are fantastic; this will be great for everyone.

I didn't know large databases used Terabits; must be a DB2 thing.

Obviously, you never worked with a multiterabyte SQL Server database running 64cores and 256GB of memory.

Oh and Alan Parsons if it's running AIX, PowerX and DB2, you've wasted your time and money locking youself into the stone age.

Lemme guess, you store all you music in .wav versus .mp3, too? Let it go, dude, x86 and the Os's that run on it won a long time ago.