Western Digital hard drive WD1000JB Special Edition 8MB Cache
I've lost a fair few hard drives over the years, most recently earlier this week when a 20GB Fujitsu decided it was time to rest its platters and head for that giant data dump in the sky. Whether by luck or by design only three of the many hard drives I've owned made it though the 90s and have almost become fond friends. They have several things in common, they're all small by today's standards, they're all slow by today's standards and they all feature the name "Western Digital". Now that might sound like a serious case of sucking up but it's pure fact and although I can't say for certain that this is because of superior engineering it certainly suggests they're not victims of poor engineering. Of course back in those days life was a lot easier for a hard drive. Data was less densely packed, rotation speeds and heat were significantly lower and they tended not to be thrashed for hours on end performing video editing or playing heavy duty games, but they still failed and I have probably had to bid a fond farewell to at least half a dozen sub-2GB drives and probably the same number of more modern 2GB+ drives. I've got to admit this trend worries me, especially as capacity grows. If I'm going to be tempted by manufacturers to trust possibly 100GB of precious data to a hard drive I want to be pretty damned sure that everything possible has been done to safeguard that data, and I don't want to have to spend days backing everything up onto 150+ CDs just so I can sleep at night.
For the record when I contacted Fujitsu about the 20GB unit (only 10 months old) I was told their warranty is solely with the supplier, not with them. The supplier went bust about 6 months after I bought the drive so it's now nestled at the bottom of the trash can. Buyer beware!
Alright so I've determined
that I tend to trust WD drives a little more than the others simply because of
good experiences in the past. You may feel the same way about Maxtor or Seagate
or IBM (possibly), so why choose WD now?
Anyway, enough rambling, let's take a look at what we're all here for, the Western Digital Caviar WD1000JB "Special Edition". If you've read reviews elsewhere you'll know that the standard 2MB WD1000 was not exactly lacking when it came to speed and capacity, but the new "Special Edition" version now features four times the amount of cache with a full 8MB on offer. This amount of cache has only been seen previously on SCSI drives along with the usually horrendous price tag, but the benefits such a large cache on a traditional IDE drive should in theory mean the WS1000JB is a superb choice for running in relatively cheap RAID arrays. Whether its benefits in other situations are worth the extra moolah remains to be seen.
In The Box :
The first thing worth mentioning is the way the drive is packed. By using two molded plastic ends the drive is suspended well away from the walls of the box. The plastic is also fairly thin which will undoubtedly help it to soak up the bumps and knocks received as your friendly delivery guy drop-kicks it into the back of his vehicle.
Despite the fact that the supplied installation instructions are very thorough there's a quick install guide printed on the rear of the box along with some of the specs. The box is your fairly standard fayre although it does stand out from the crowd thanks to the shiny silver "Special Edition" stickers that have applied to the front (see top of page). There's no bold claims other than this, no talk of ceramic spindle bearings or fluid dynamic bearings, no talk of GMR heads or high-tech vibration damping, in fact there's very little to get your juices flowing and decide you simply have to own this hard drive. The OEM roots are clear but it's time for WD to learn the art of hype.
Supplied alongside the drive, other than the warrantee registration card and installation gubbins are a cream coloured IDE cable (different!), mounting screws and a floppy disk containing WD's excellent Data Lifeguard Tools. So what's Data Lifeguard? I'll let WD tell you :
Data Lifeguard Hardware Features
Data Lifeguard Tools Software Utilities
For the newbie, the Data Lifeguard software is a life saver (excuse the pun). You simply place it in your floppy and allow your system to boot from it then choose from a set of options that will automatically partition and format your new drive ready to go. There's even a clone function that allows you to copy the contents of your old drive across to the new one. Combine this with the comprehensive set of hardware data monitoring functions and you begin to feel safer that your data is in good hands.
If you've installed one hard drive you've installed them all. One strange problem I did run into though was that it refused to appear in XP when set to Master with a Maxtor ATA-133 slaved behind it. The only way I could get it to show up in XP was to set both drives to "Cable Select" and make sure the WD was placed on the last (master) connector on the IDE cable. No great problem but one I've not come across before.
The WD1000JB comes with four mounting screws so no need to dive headlong into your "bits" drawer. The rest is the usual case of partitioning and formatting that goes along with every HDD install, either manually through DOS or by using the Data Lifeguard floppy. Of course partitioning and formatting a 100GB drive takes some serious time, probably around three quarters of an hour in total so grab a coffee and a good book before you start.
If there's one thing that becomes clear when you're testing a hard drive it's that the industry sorely needs a modern, thorough and reliable HDD benchmark. Some of the benchmark results I got differed by huge margins, others threw up a page full of confusing or pointless numbers that wouldn't really serve the purpose of giving a clear and easy to understand performance index that would assists in making an informed buying decision. Although I've published some of the easier to understand results below but this is one of those occasions where the best feel for performance comes from using the thing for a while and I'll come to that "seat of the pants" perception later.
Test System :
Epox 8K3A+ (KT333)
SiSoft Sandra File System Benchmark :
Sandra's file system benchmark is designed to evaluate all aspects of your drives and controllers. It's a synthetic benchmark and it's not particularly thorough when the result is used alone but as a comparison run for two drives on the same system it does give a fairly useful indicator. Tests performed are :
Testing the Maxtor it was clear that performance was pretty impressive. It bested the index score by a fair margin.
However when the WD1000JB was tested it simply blew both the index score and the Maxtor's score clean out of the water. To pull ahead of the Maxtor drive by a clear 15% is a genuinely impressive result for two drives that share so many similar characteristics on paper. Let's chalk that one up to WD!
HDDSpeed 2.0 :
HDDSpeed is a DOS run benchmark that offers a variety of test parameters designed to test HDD performance. Several were run but only two are used here. In the linear read tests I was slightly surprised to see the Maxtor pull ahead with an average read that was a good 0.5MB/sec higher than the WD. I wasn't able to find the Aerial density figures for the WD100JB so I can't comment on whether or not this is a factor.
Linear Reads :
Cache Speeds :
Not surprisingly the cache speeds were pretty much identical with both drives topping out here at 7MB/sec. The WD may have more cache but that doesn't mean it's any faster, in fact with more cache the tendency would be for latencies to increase as it takes four times longer to scan 8MB of cache for data than it does 2MB.
HDTach 2.61 :
HDTach is one of the more popular hard drive benchmarks though like the others it has its failings. I'd already partitioned and formatted the WD which meant no write speed tests unfortunately and with a rake of benchies to run there was no way I was going to dump the active partition just to suit HDTach.
Here we can once again see the WD flex its muscles. Access times are virtually unchanged over the rest of WD's range of drives at 13.3ms and while this is a perfectly respectable figure it can't compete with Maxtor's excellent 11.7ms. This however is Maxtor's only win as it then goes on to be soundly beaten on minimum, maximum and average speed results.
Intel's IOMeter is another benchmark that comes with a bewildering array of settings and configurations. Some sites have built their own access patterns for use with IOMeter but I stuck with the Intel generated options and rather than show the results from around 9 different configurations I decided to stick to a single one which I figured was a good compromise and should be very easy to set up if you want to run it at home. For this one I simply used the "default" configuration but bumped up the Transfer Request Size to 128k. For the record the WD showed a handsome lead in every test run and the results below are pretty indicative of the others.
With a total I/O/sec figure
of 91.33 the WD1000JB holds a massive 17% advantage over the Maxtor, in fact
the MB/sec throughput lead also goes to WD by a similar 17% margin.
Heat and Noise Creation :
7200RPM drives tend to get hot but it was good to see that the WD1000JB remained relatively cool compared to other similar drives and a full three degrees cooler than the Maxtor. I have no facility for measuring sound levels other than the ones on either side of my head but they were certainly nothing to get upset about. I've heard quieter but I've also heard louder suggesting a kind of "middle of the pack" position.
Larger hard drives need more cache and WD have set the standard with the Caviar "Special Edition" series. That's not to say that an increased cache will bring massive benefits for everything you do, in fact for the vast majority of uses WD's standard 2MB drive (WD1000BB) offers comparable performance. What makes the special edition model a worthwhile investment is that it costs less than $20 more than the standard model (as seen HERE)
Western Digital took a slower, more steady approach and were quite late to the large capacity and high performance hard drive sector but it's looking like it paid off. Compared to the problems faced by some of the big names who jumped in with both feet only to find performance and reliability was a major problem for them the WD1000JB seems to be a much more robust drive with a superior lick of speed. Let's hope I'm still saying that in 12 or 18 months time.
For mere mortals like me there's no quick and easy way to fully test hard drive performance under a full range of situations and I wish somebody would step in and create a quality benchmark that takes out the guesswork and compiles some valid, accurate and easy to understand performance figures that everyone can make sense of at a glance. Even the dedicated storage related sites have to constantly shift the goal posts in order to keep on top of developments and quirks in drive design and performance attributes so it makes it almost impossible for sites like us to really get our teeth into what's on offer.
What I can say with confidence however is that the Caviar Special Edition WD1000JB is the fastest hard drive to have ever graced our test bench and impressive though the Maxtor D740X may be it simply gets outclassed in just about every test with the exception of access times and linear reads. I mentioned higher up the page that sometimes the best performance indicators come from using a product and the WD1000JB certainly feels fast. From DV editing to file shuffling and database work you get the definite impression that the WD1000JB is barely breaking a sweat, always a good feeling when you're trusting your crucial data to it.
I can't comment on long term reliability simply because I can't recreate 2 years workload in a couple of days testing but certainly based on past experiences I'm expecting big things. WD did have a few problems with faulty chips in some drives back in September 99 but they immediately set in motion a full and aggressive product recall scheme and the problem was swiftly dealt with. Add to this a three year warranty and WD have most of the bases covered.
So do you need an 8MB cache? Probably not but it does help in certain situations and like graphics cards sometimes even a small advantage can be worth having, particularly when as with the WD1000JB the price premium isn't too huge. Whether or not the WD1000JB remains a limited edition product depends very much on the sales figures which are being carefully monitored but certainly at this point in time WD don't intend to produce this drive in huge numbers. Speed and exclusivity! What better reason to buy?
So far I'm impressed. I'm
not sure the claims on the front of the box regarding the WD1000JB offering
near 15kRPM SCSI drive performance is quite accurate but it's not too far away
either, you're certainly going to get performance equal to and in many cases
beyond those from a good 10kRPM unit. The price may seem a little high but it
doesn't take too much searching to find a much healthier street price.
Western Digital 1000JB SE 100gb HDD Review
15. August 2002
Before I begin I'd like to thank Sterling from Monarch Computer for sending over this drive.
I guess it was gonna happen sooner or later. My 30gb IBM 75GXP died at a very inopportune moment. I was in the process of putting together a "new" system for my mom (only a Duron, but better than the P2 300 rig she had earlier). The older one ran off a 4gb Quantum drive that was well under 5400rpm, probably 4200. This was totally unacceptable for what I was building. Furthermore, my main system (the one where I keep all my files, website or not) was running off an IBM 60GXP which was also infamous for this whole failure rate business. As you can imagine, I wasn't too thrilled. I immediately backed up all my important data onto CD for the first time ever (everyone needs a wake-up call) and went in search of a drive.
After looking at the possible alternatives I ran into several choices. The Maxtor D740X, the IBM 120GXP and the Western Digital JB1000 SE. I needed a big drive as I was planning on setting up a Win2k/WinXP dual boot. In addition, I'm not big on having the latest and greatest so my motherboard is only an Abit KG7-RAID (AMD 760 chipset) running an AXP 1800+. While Maxtor's D740X is now available in 80gb flavors, at the time the only widespread models were the 20 and 40gb versions. Furthermore, I would need a separate IDE controller to take advantage of the extra "33mb/s" data throughput. No thank you.
The IBM 120GXP, well, I decided to stay away from anything IBM at least a while. I can't rely on one company completely. Although failure rates with this drive are at a minimum and the 40gb/platter performance is very impressive (on par in a lot benchmarks with the drive I'm reviewing today) I was still a bit skeptical, especially after reading this controversial article. After reading nothing but praise about Western Digital's "accidental" drive, I decided to give it a spin. So my bud Sterling over-nighted me a spankin' new 1000JB SE. The drive is available in 80gb (800JB), 100gb (1000JB) and 120gb (1200JB) versions.
Before I go on I should mention that I didn't compare this drive to either the 120GXP or the D740X because I have neither. If you want to flame me please attach a drive with the email. Therefore, I'm not going to say "this is the best drive" because that would be blatantly disregarding the other competitors. Instead, I'd like to show you how fast the drive itself is and how it compares to a "regular" 7200rpm drive, the IBM 60GXP. With that said, let's begin.
I got an OEM version. That is, the bare drive sealed in an anti-static bag. The retail kit contains the drive, a utility floppy, an ATA-100 cable, a user's manual, and a few screws. That's ok, I have all that stuff. The most appealing thing about this drive is the 8mb cache. How did this come about? Well, it was actually a sort of accident. Here's an excerpt pulled from x-bit labs' 1200JB review:
By an oversight of one of WD's software developers, the hard drive DSP-processor could now access not only 2MB, but the entire 8MB memory. And which is also unbelievable, the HDD with this "crazy" DSP-processor turned out evidently faster than its counterparts featuring memory modules of lower capacity.
Much like the rewrite speeds of CD-RWs, over the past few years the cache went from 2mb to an amazing ... 2mb. Why didn't anyone use 4mb or 6mb? Don't ask me. All I know it Western Digital just busted out with an 8mb cache. In addition, this series of drives is based on 40gb platters much like the 120GXP and the D740X.
The drive itself it extremely well built. The upper portion is standard: metal casing, sticker, air hole. The underside, however, is what surprised me. For one thing, the circuitry is reversed. That is, all the chips and most of the traces are on the inner side of the drive, therefore preventing possible damage. There's also a layer of foam in-between to further lessen vibrations, add impact cushioning and protect the drive.
I didn't particularly feel like removing the PCB
to see what's on the other side (after all, this was a rush-delivered drive that
is now my main system's powerhouse). There isn't much interesting aside from an
8mb Nanya memory module which happens to be the cache. The 1000JB SE also seems
a bit quieter during operating than the only other drive I had available, the
60GXP. I wish it were as quiet as Seagate's Barracuda IV but I guess you can't
have it all. For one thing, it doesn't hiss as loud and neither does it
"crackle" when being stressed. As far as heat goes it stays about as cool as
most drives. Now for some specs:
Ok so it's not exactly 100gb and the more you partition the more you lose. Since I'm using dual boot I opted for 5gb Win2k OS, 5gb WinXP OS, 60gb WinXP files, and 23.1gb Win2k files. That's a total of 93.1gb. I feel cheated o_0. Anyhow, the seek and read times are pretty average as the larger cache doesn't have much of a say in that. So what does an 8mb buffer do? In the words of Western Digital:
The Special Edition hard drive provides a higher percentage of cache hits and significantly faster time to data than industry-standard 2 MB versions. Within the complex operations of hard drives, the buffer is used to hold the results of recent reads from the disk, and also to "pre-fetch" information that is likely to be requested in the near future. The use of an increased buffer improves the performance by reducing the number of physical accesses to the disk and allowing data to stream from the disk uninterrupted by mechanical operations.
And a bit more about Data Lifeguard:
Data Lifeguard is an expanded set of data protection tools that includes shock protection, and environmental protection system, and embedded error detection and repair features that automatically find, isolate, and repair problem areas that may develop over the extended use of the hard drive. With these enhanced data reliability features, the drive can perform more accurate monitoring, error repair, and deliver exceptional data security.
Data Lifeguard Tools™ is a set of software utilities that work in conjunction with the embedded Data Lifeguard features to make hard drive installation, drive management diagnostics, and repair simple and worry-free.
Shock Guard™ provides outstanding improvements in shock and vibration protection. Shock Guard allows instantaneous data protection at high shock values to achieve leading shock specifications.
The sticker on the drive indicats pretty much all you want to know, from parameters to the capacity to various jumper settings. In the upper right corner is a warning against covering up any of the air holes used for passively cooling the drive. Now I didn't toss this somewhat expensive drive around the place to see if it really was fragile but compared to several other drives I've seen the Western Digital is a lot sturdier, mostly due to the reversed PCB and foam cushioning.
The rear of the drive is nothing special. There's the 4-pin molex, a set of jumpers and the 40 pins. By default it's set to cable select which is what I left it as. This may or may not work for some systems so one of the other options may be in order (single neutral, single standard, dual master, or dual slave). One thing I particularly like about the 4-pin connector has to do with the inverted PCB. Because of this the molex is "locked" in place and doesn't wiggle when you plug in/remove the wires. On most other drives yanking the cord out a bit hard can break off the weakly attached connector.
Size wise the WD is a bit taller than the IBM drive. This could have something to do with a thicker plate or the extra space in-between the PCB and the "real" underside (where the cushioning is). Once again I regret not having either of the other two high-end IDE drives for comparison but we'll have to make do. I know it's not much but at least it's something. For testing I used Windows 2000 with all the current updates. Unfortunately I still don't have some of the other well-known programs but if you know of any reliable benchmarking utilities please feel free to drop me a line. For testing I used:
First up is Sandra:
SiSoft Sandra is a somewhat controversial benchmarking utility. It's come a long way but there's still a lot left to be done. The reason I say this is because every time you run a benchmark (no matter what you're stressing) you'll get slightly different results. Sometimes it's within 1-2% while other times it's quite noticeable. I only ran it once. The WD outperforms the IBM by quite a bit and is on par with the reference 120gb 7200rpm drive while falling far behind a 15000rpm SCSI. IBM's downfall here is due to the Western Digital's larger platter density and greater cache. What's really interesting, however, is that the 60GXP's buffered and random reads are faster, and so is the average access time. However, during the sequential read and write (aka sustained) the WD comes out victorious by a significant margin.
HD Tach reported similar results. Both drives have nearly identical access times. For some reason (this could be glitch but then again it might be real) the IBM's read burst speed is off the charts with no assigned value. I ran the test several times and got nearly identical results. However, the maximum, minimum and average read speeds of the 60GXP are significantly slower; about 2/3 of the Western Digital's. Another interesting observation is the CPU utilization; the WD loads the processor about twice as much as the IBM. Could this be a sign of the 8mb cache at work? Possibly. Since larger cache size theoretically = less physical accesses, the CPU may be working more than usual to feed information on time.
Western Digital 1000JB SE
Western Digital 1000JB SE
PC Mark 2002 continues the trend. As you can see all write, read and copy speeds are a good deal faster with the Western Digital drive. I guess I should repeat this again: I'm not bashing the IBM 60GXP. It's not in the same league but it's the only drive I had. Furthermore, most drives on the market are about the speed of the 60GXP with few exceptions. Hopefully in the near future I can acquire something faster for a better comparison.
As for the drive itself, Western Digital sure made a winner. It's fast, cool, quiet, reliable, and several months since introduction, cheap. Prices are finally down about $100 and comparable to that of the 120GXP. And given IBM's controversial nature with previous drives and now this one (although failures aren't as common), I'd personally get the WD just to be on the safe side. Then again, I already have one. I'd like to thank Sterling from Monarch Computer for sending over this drive and in interested you can get yours here. If you have any questions or comments let me know here or post in the forums or both. Thanks for reading and enjoy the site!
Copyright © 2002