Magneto-optical disks can be attractive storage devices for many applications. If you have massive amounts of data that you want to store, and if you tend to access large blocks of data sequentially (if you’re reading or writing large files), they can be extremely cost-effective. However, due to its slower seek time, today’s magneto-optical drive usually isn’t seen as a suitable replacement for a Winchester-style (i.e. normal) magnetic disk.
Most magneto-optical drives use a fairly massive read/write head, which contains both magnetic and optical components. The head assembly weighs more than the arm-and-head assembly on a Winchester drive, and it can’t move across the disk as quickly. As a result, seek times for a typical magneto-optical drive are often several times slower than seeks on a Winchester drive of a similar capacity.
In addition, most magneto-optical drives write data more slowly than they read it, because the laser must make two or more passes over the medium – one pass to erase the old bits, one pass to write the new ones, and (in some cases) one pass to verify the data. Once again, these drives are slower than Winchesters of similar capacity.
Something new? — The Pinnacle Micro PMO-650 has a two-page ad in the 25-May-92 issue of MacWEEK which claims that this mechanism is a bird of a different color because its speed compares to that of a reasonably fast hard disk. However, further reading of the ad suggested that there is a bit of fast talking and clever verbiage taking place, and that the ad is at best misleading in its claims.
The ad cites a "19 millisecond effective access time." There’s a footnote bullet which states:
Test Results performed with a 1/3 stroke plus latency over a 50 MB band width. A 1/3 stroke plus latency is the standard of measurement in the optical storage industry.
Well, this may be true, but if you try to compare this "19 millisecond" figure to the equivalent figure for a similarly-sized Winchester hard disk, you will be comparing apples and oranges and may be disappointed.
The catch is the "50 MB band width" clause. This means that Pinnacle has benchmarked their drive by performing seeks over a 50 MB band on the medium, which is only 50 MB out of the 325 MB on one side of the platter! In other words, Pinnacle Micro has based their "1/3 stroke" on the assumption that a "full stroke" covers less than a sixth of the data area on the platter!
This is particularly significant, because this drive uses a split optic design. Drives of this sort can perform short seeks quite quickly (they usually move a tracking mirror using a light voice-coil mechanism) but slow down on long seeks (the entire laser assembly must be moved). By restricting their seek-time tests to a 50 MB band, Pinnacle may have skewed the results of the test in their favor since a larger percentage of the seeks can be performed using the fast-seek mechanism than would be possible in a full-disk-surface seek test.
Checking out the facts — I called Pinnacle’s 800 number, asked for information about the drive, and spoke with a representative for about five minutes. I asked a number of questions and got the following responses (summarized, not quoted exactly):
- Am I correct in assuming that the seek tests are being performed across only one 50 MB band out of the 325 MB on the platter surface? [Yes]
- Most Winchester disks have their "effective access time" based on a typical mix of seeks over the full disk capacity, do they not? [Yes, I believe so.]
- Then I can’t really make an apples-and-apples comparison between a Winchester disk and the PMO-650 based on your quoted access time, can I? [Well, you can’t really compare the two technologies, because hard disks have multiple heads and optical drives have only one.]
- But that’s the kind of comparison your ad tries to make, isn’t it? [Yes]
- The ad notes that "A 1/3 stroke plus latency is the standard of measurement in the optical storage industry," and I understand that this sort of measurement is fairly typical in the Winchester disk industry as well. [Yes, I believe that’s correct]
- Do you know of any other vendors of optical disks who quote an average seek time based on a test which limits the full stroke to only a small fraction of the platter capacity? [Offhand, no, I can’t think of any.]
Another opinion — A couple of weeks after this, I spoke with a representative of a company which makes high-performance caching controllers for magneto-optical disks. I described the Pinnacle benchmark as it had been advertised, and asked him what he thought of it. His response was to the effect of "That’s nonsense." (he used a much stronger word than "nonsense") He told me that he considered Pinnacle’s claim to be harmful to the magneto-optical disk industry, because it created false expectations about the performance of that sort of drive. He hadn’t heard of any other vendor which printed benchmarks based on a 50 MB band of the disk and was skeptical about the whole concept of trying to compare Winchester and magneto-optical disks through any sort of standardized benchmark.
It’s actually a pretty hot drive — It may sound from all of this as if I’m entirely negative about the PMO-650 drive. Actually, I’m not. There are some other facts about the drive which may – and probably do – allow it to outperform similar mechanisms by a substantial margin.
The drive controller has a large cache – 4 MB. According to Pinnacle representatives, the cache management firmware is quite sophisticated – it supports read-ahead, write-behind, and a number of different media-access algorithms. As a result, the drive can frequently "service" one request from the host computer while writing out the data from previous requests to the media in an orderly fashion. In addition, the drive has a respectable sustained transfer rate – 1.3 MB/second on a Mac IIfx, 1.5 MB/second on a Quadra 900. That’s quite a bit faster than either of the two Winchester disks I own.
My conclusions? — On the positive side: for an optical mechanism, this drive is probably faster than most. If you’re doing the sorts of jobs for which magneto-optical disks are often suggested – for example, working with big 24-bit TIFF files – you’d probably be quite happy with this drive.
On the negative side: if you’re expecting this drive to behave as well as a good Winchester, under a broad set of conditions, you may be disappointed. Test it yourself, if you’re considering it for a performance-intensive application and don’t trust Pinnacle’s quoted performance figures unless you’re satisfied that their test conditions match up with your intended usage of the disk.
Caveat emptor, folks. There are lies, damned lies, and benchmarks. There is also advertising. In my opinion, Pinnacle did a bad thing. They chose to print an ad which makes a misleading comparison, and they made statements in the ad which are unsupported by the facts. Somebody in their organization, their advertising firm, or both, deserves to get sacked for trying to pull a scam like this.
It’s a shame, really. It looks as if Pinnacle Micro has developed a nice product, and has good reason to be proud of the PMO-650. I’m not sure I’d be willing to buy anything from Pinnacle Micro, though, as I’m not confident that I can trust them.
Pinnacle Micro — 800/553-7070 — 714/727-3300