Characteristics Of 4K Sector Size (Advanced Format)
WD adds to the WD10EARS drive label general, straightforward instructions on how to set up the drive for maximum performance. You simply use the WD Align software to achieve maximum performance on Windows XP systems with multiple partitions. Or, if partitions were cloned, or you set jumpers 7-8 prior to installing Windows XP on a single WD10EARS partition. All other operating systems, meaning Windows Vista and Windows 7, do not require any user action.
This information already tells us something: apparently, the two latest Windows editions are ready to work with 4KB sector size, while Windows XP is not. In addition, according to WD’s instructions, there seems to be something that needs to be aligned under certain conditions. This refers to logical versus physical sectors. All drives that utilize 4KB sector size still emulate 512 byte sectors on the interface side for compatibility reasons, as applications will continue to use 512 byte sectors for quite a while. Therefore, the new 4KB physical blocks store eight 512 byte logical blocks. This is where issues may crop up.
The Performance Trap
If 512 byte data has to be written across two physical 4KB blocks, the hard drive will have to read the 4KB blocks that are affected, introduce the modifications, and write them back onto the drive. This process is referred to as read-modify-write (RMW). While this doesn’t hurt much if it happens here and there, it becomes a significant issue if the alignment of eight 512 byte sectors into only one physical 4KB block is wrong. Therefore WD offers either its Alignment software (information and downloads can be found here), which rearranges all data on a hard drive to fit the logical 512 byte sectors into the physical 4KB sectors.
Windows XP vs. Vista and 7
Windows Vista and Windows 7 are sector-aware, meaning they will automatically align 512 byte sectors with the hard drive’s physical sectors starting at LBA block 2048. Windows XP didn’t have this capability, and since the OS creates the first partition at LBA sector number 63, all following sectors will be misaligned, causing a noticeable performance impact. Each write process now causes a RMW that introduces additional latency due to the required additional rotation of the hard drive.
Linux?
Linux fdisk makes partitioning a piece of cake for 4K hard drives. The parameter –u allows defining an absolute sector number. Any sector that is divisible by eight will work fine.
WD’s Alignment Solutions
Western Digital currently offers two tweaks to avoid the performance issues. The first one is the Alignment software that simply relocates all available partitions to make sure they start at a fresh physical 4KB sector. The second option is the mysterious jumper 7-8 on the hard drive, which we don’t recommend using. It's a pretty reckless hardware hack on the partition table, as it simply increments the effective LBA sector number by one. As a result, Windows XP, which creates its first partition at LBA 63, will be writing onto logical block 64 to align with the drive’s 4KB sector size. Unfortunately, all your data will be gone if you operate the drive without the jumper, while the Alignment tool is not harmful. All you need is a little patience until the drive relocates all partitions to align logical into physical sectors. The Alignment tool took a bit more than an hour to relocate a roughly 20GB test installation with all important drivers and some benchmarks installed.
I wouldnt have it as a boot drive (due to its write speed) but as a data drive, it would be quite nice.
7-11% extra capacity for worse write speed, i would take that. seems like i may have found my next storage drive as soon as they offer higher capacities of this.
Soooo... any chance these are compatible with WHS systems? I didn't read the entire article but I saw no mention of it on the benchmarks and what I did read.
Don't really care.
More interested to know when I can pick up a nic 4TB drive though. Hopefully rocking right next to my 256GB Gen3 x18-m. =)
It is compatible with any system (including DOS) - it's just that performance won't be good on systems whose partitioner is unaware that Logical Block Addressing (which works with 512 Kb sectors even on 4K sector drives) don't match the actual drive's layout.
It also means that if you use a partitioning tool that is sector aware, you can install any system optimally.
Case in point: you're using a Linux LiveCD to partition your whole system. You create a primary NTFS partition. You create all the other partitions you want, of any kind you prefer.
Then you start installing Windows Home Server on your created NTFS partition. Done.
As long as you don't try to use funky NTFS sector sizes (NTFS defaults to 4K cluster size, but can go as low as 512 bytes - this often happens on small FAT32 partitions converted to NTFS), you should be OK.
In essence:
- create your partitions on a modern OS (at MS, anything that came out after 2006: Vista, 2008, Seven; for UNICes, any - that includes Linux, OS X, BSD, Solaris)
- format these partitions with 4K (or multiples of 4K) cluster/inode/sector size. On Windows systems, you can even use Windows 2000, as it also defaulted to 4K sector sizes for NTFS on new partitions. You should forget about FAT32, but even that should work with cluster sizes >= 4K. Cloned partitions should have >= 4K block sizes too.
- install as usual.
I'm not that sure about the 10% capacity increase in practice: won't any file, even a 1-byte one, now take at least 4kB on disk ?
Manufacturers get to advertise 10% capacity, users get x% extra waste. It would be nice to look at a typical Win7 system, at compute whether users actually gain or lose HD space.
PLus, what's the impact on SSDs ?
The majority of these new technology drives i.e. GP/LP low energy models and ultra high capacity models, above 1TB, and especially 2TB have such appalingly high fault and failure rates as to make them an insane choice for anyone especially since their main use is data storage. If you're running a RAID 1 array that allows hot swapping and really need massive capacity you might choose to risk it, but for a desktop to user who wants to store their photos, videos, tunes etc, they are extremely high risk.
Check out the ratings given by users on newegg to see an illustration.
Drives up to 1TB typically score 80% statisfaction rates (4/5 stars). Drives above 1.5TB are at 30% with up to 50% of people scoring them extremely low (1/2 stars) and citing DOAs, failure after 1-3 months, data loss due to poor firmware or incompatibility with a variety of controllers... Especially telling are the users who order multiple drives for larger systems - they report between 1 in 4 drives to 2 in 4 drives being faulty.
I have a friend who works at the largest manufacturer and when I asked him about this he basically declined to answer. The companies know their yields are terrible but they have no choice since people want those high capacity drives.
I would personally love 4x2TB GP drives for my NAS but I will not risk it until 2 more generations of the technology come out.
I'm not that sure about the 10% capacity increase in practice: won't any file, even a 1-byte one, now take at least 4kB on disk ?Manufacturers get to advertise 10% capacity, users get x% extra waste. It would be nice to look at a typical Win7 system, at compute whether users actually gain or lose HD space.PLus, what's the impact on SSDs ?
At default, WinXP does format NTFS partitions with a 4k cluster size. So you're not actually losing anything new with 4k physical blocks.
@obarthelemy: you should learn to read.
No, the waste you're talking about is at the file system level - and is already the case in current drives, since file systems aren't impacted by the change.
As I said in my previous comment, and as is said in the article (but indirectly), performance is lost when there is an alignment mismatch between LBA addressing and physical addressing.
Up till now, LBA addressing was related to the physical layout: LBA used 512 bytes logical sectors, which the drive's eletronics matched against the platter's physical blocks - which had a 512 bytes payload+spacer+CRC.
On top of that, were placed file systems - and currently, most file systems use a 4K cluster (FAT) or sector (NTFS) or inode (ext2/3/4) size.
Meaning that a file system logical block commonly took 8 LBA sectors, which related directly to 8 physical sectors.
With these new drives, you have a file system block taking 8 LBA sectors, which relate to a single physical sector.
But that is the best case. What happens if the file system starts at a non multiple of 8 LBA sector? Well, a file system's block will ride on top of 2 material blocks, and the disk will have to read both sectors, and piece together the actual data needed - requiring twice the number of reads.
SSDs don't work the same way. If you read the article, you'll see that this is a DESIGN FEATURE of hard disk drives; SSDs are NOT hard disk drives, thus they are NOT concerned by this problem. Well, not the same way at the very least.
http://www.youtube.com/watch?v=473kNyO9jUM
PS those with slim cases need not to apply.
http://www.youtube.com/watch?v=RBg [...] re=related
When drives are big.
The majority of these new technology drives i.e. GP/LP low energy models and ultra high capacity models, above 1TB, and especially 2TB have such appalingly high fault and failure rates as to make them an insane choice for anyone especially since their main use is data storage.~~ for a desktop to user who wants to store their photos, videos, tunes etc, they are extremely high risk. Check out the ratings given by users on newegg to see an illustration.
I agree. The loss of 2TB of data would be horrible. But I wonder how many of these 2TB users are just storing lots of porn and movies?
I am able to fit most of my needs onto 500GB. Store excess data or backups onto 1TB drives. Also, remember last year when Seagate had the problems with their 1.5TB drives failing? But all this high capacity filters down to smaller drives by reducing their platters.
I remember buying a 160GB drive a few years ago and its half as thin as a normal 3.5" drive. The last 500GB drive I bought for a customer build is also very slim. It shows that its the same tech used in their 1TB drives, just less platters. Eventually all HDs will be like this, where they can fit 2TB of data onto a single platter.
With more platters, comes more heat and higher chance of failure with the extra heads, etc.
Most reliable are 1 if not 2 platters total for a drive... but I'm afraid as the density increases, so does the failure. SSDs are still advancing but still a good 3~4 years behind in price/storage ratios.
BTW: People should check out the noise levels. I've told people that I found Seagates to be pretty much silent, the chart shows that. I built a system with a WD drive last year, I couldn't stand the noise in an otherwise silent system and exchanged it for a Seagate.
http://www.youtube.com/watch?v=473 [...] atch?v=RBg [...] re=relatedWhen drives are big. PS those with slim cases need not to apply.
Blah... those are still SMALL compared to some IBM server drives that sound like JET engines when powering up, usually taking about 3-5 minutes to reach full speed. There were the size of dishwashers.
Heres a cool video, shows the insides of the drive working:
http://www.youtube.com/watch?v=Gb1 [...] re=related
Notice, with such huge areas to work with, its not as suspectable to damage as todays drives, but then again it only holds 1/1000th of todays $10 flash drives and weighs about 10lbs more.
For sick fun: http://www.youtube.com/watch?v=qWV [...] xt_from=PL
BTW: No youtube videos on those old IBMs... that was back in the days before cheap video hardware and the internet.
Soooo... any chance these are compatible with WHS systems? I didn't read the entire article but I saw no mention of it on the benchmarks and what I did read.
Server 2003, which WHS is based on, is similar to XP/2000, so you'd have to run the alignment utility. I wouldn't use it as the OS drive, but I have heard it is OK for the data pool.
I've thought about using them too, especially when I see them on sale, but always am worried about how well they would work in WHS.
A nice incremental improvement.
The 10% increase in track capacity will not translate into 10% more drive capacity. The reason is that drives will still be made in 500gb/1tb/2tb... capacities. Costs will not change much either because 10% is not enough to eliminate a platter.
What WILL change though is a 10% increase in data transfer rate. That is because 10% more capacity can pass under the R/W heads in the same revolution. That's good.
What if you're using two of these drives in hardware RAID1 in XP. Should the utility be run or is it not necessary?
Wait a sec, wouldn't 4KB sectors help given the OS data benches done with SSDs are accurate?
A bit more performance... this doesn't mean much.
But my question is if I use NTFS 64k cluster, what is the impact on 512k HD clusters or 4k HD clusters ?
This is the real question they should have answer.
Bad artical!
To much focus on the extra storage, that might get wasted if you have small files!
...and almost not a word on what is the most likely problem, even worse failrate on reading. YES modern disk drives do fail on read! So often that a rebuild of a raid on large disk can be at risk!
Com'on Tomshardware! This article should have been saying to the industry. BAD concept! BAD idea!
Heres a cool video, shows the insides of the drive working:
http://www.youtube.com/watch?v=Gb1 [...] re=related
That's awesome, I actually have one of those exact same Micropolis drives at home. It still works too! I think it is like 660Mb or something like that.
I wouldnt have it as a boot drive (due to its write speed) but as a data drive, it would be quite nice. 7-11% extra capacity for worse write speed, i would take that. seems like i may have found my next storage drive as soon as they offer higher capacities of this.
Problem is, they are NOT giving you any extra storage capacity, its still a 1, 1.5, or 2TB drive with the same number of platters. If they gave you a 1.1, a 1.65, or a 2.2TB storage capacity, for the same price as the old tech, then it would be worthwhile.
These drives are just a test vehicle for the new technology, and don't have any real benefit over the previous drives from a consumer standpoint. Its a step in the right direction, but personally I'll hold off for now.