XP or Vista…which one’s better for SSDs?
We’re pretty passionate around here about engineering SSDs that get the highest performance possible. Part of that engineering effort is within the devices themselves—ensuring that our SSDs live up to the potential that our NAND flash offers. That said, there are many variables outside of the SSD that impact performance—like the operating system or hardware interface. After all, many of today’s SSDs are being asked to live in a world designed for HDDs.
The SSD group at Micron has done extensive research into the behavior of Windows XP and Windows Vista under different conditions. To gather data, the SATA interface was analyzed and data captured while installing, booting, shutting down, and running Office productivity applications on both operating systems.
The results of this research show that XP and Vista do indeed handle their interfaces to an SSD differently. The main difference is XP does not align the data in the best way for an SSD. Vista, on the other hand, does align the data. These findings differ from other claims we’ve seen that Vista is somehow worse for an SSD. In fact Vista is better for an SSD (incidentally, so is Windows Mojave). The example below illustrates the difference… By aligning the data and not partially filling NAND pages, performance is optimized for the SSD.
Click to Enlarge...
Another key difference between Vista and XP is that Vista enables background drive defragmentation by default. Defragmentation is totally unnecessary with an SSD and actually wears the drive out more rapidly. However, most OEMs will disable background defrag on systems that ship with SSDs and Vista. So—if you happen to be buying a new SSD for the holidays, be sure defrag is off when you upgrade. You don’t want to hold performance back by making your new SSD live in the HDD days-of-old.
No related posts.
Related posts brought to you by Yet Another Related Posts Plugin.
7 Comments
GullLars on December 4th, 2008
Great article here. I was actually surprised by your findings. However i have a question: Even if vista alligns the writes better, wich operating system does the most background writes? And how much background write is done in real life desktop enviroments? It would be great to see some estimate numbers of MB/s (or KB/s), write block sizes, IOPS, and how much of this is sequential vs random.
Jason Tan, the information you are posting there is partially incorrect, and can be wrong for many SSDs that manage this correctly.
This storage blog already has an article on this, reffering to it as garbage collection.
This can also be seen as a sub segment of wear-leveling.
For more info on this speciffic matter you can google Intels whitepapers for their new drives, as this is one of their main focus points.
Dean on December 4th, 2008
Hello again, GullLars! So, in identically configured systems we have not noted any significant difference in background writes. Frankly, today we spend more of our time beating up these drives–trying to break them–than we do looking at background operations. While we’ve captured a lot of that information, it’s oh-so-boring … Our main interest has been to understand wear implications. Application launch and operation is a lot more interesting and is where SSDs really shine.
Dean on December 4th, 2008
Jason, a defrag utility doesn’t help an SSD free up space … today. Here’s why: Defragers move the drive contents around to put files in large contiguous blocks. In doing so, they update the location of the content and the entries in look-up tables (i.e. FAT tables) so the O/S can find the content. However, the old content location is not actually erased, it’s just not pointed to by the FAT table. To the drive, the old data is not free space. Now, in the world-of-tomorrow, Defrag utilities will know they are working on an SSD and will use some of the new SATA-8 commands to tell the drive what is free space.
Which operating system is best for SSDs? | InfoWorld | News | 2008-12-11 | By Lucas Mearian, Computerworld on December 11th, 2008
[...] Micron found that Vista and Mac OS X performed better with its SSDs than XP, according to a post on Micron’s blog site . XP does not align the data in the most efficient way for an SSD — in 4KB blocks — while Vista [...]
rudepeople on December 19th, 2008
BTW: the Linux kernel has had block device writing efficiency down since the 2.4 kernel and BSD has had it working for longer (both HAD to, as they were at some point built for embedded appliances–firewalls, routers, and such–which often ran on NAND or CF storage devices)
Don’t get me wrong, I’m not a complete Linux zealot… I feel that everyone has an OS that is functional for them, some need windows, others prefer Linux, and still more are more proficient in Mac OS… but at the same time it upsets me when corporations “forget” that Windows is NOT the only OS in the world.
David on March 25th, 2009
Interesting to note that Windows 7 has been optimized to work efficiently with SSD
Jason Tan on December 3rd, 2008
Reg. SSD defragmentation, it seems that while SSDs are immune to deterioration in read speed from file fragmentation, they are quite susceptible to poor write performance from free space (not file) fragmentation.
There is some more information available at the Diskeeper Blog which describes their approach towards SSD optimization
http://www.diskeeperblog.com/archives/2008/12/hyperfast_is_al.html#comments
(Diskeeper are the guys who rule the Windows defrag market)
According to them, file defragmentation is not required, as used to be with HDDs, but free space consolidation greatly helps to improve file write performance and drive life.
Interesting, no?