Monday, September 06, 2004

Firewire 800 vs. 400 speeds: not as large a difference as one might expect

250GB FireWire 800 Drives
The test that showed the greatest difference between FireWire 800 and FireWire 400 involved simultaneously reading and writing a large file to the same drive, which is what happens during file duplication. When duplicating a 1GB file, the FireWire 800 drives completed the task 38 percent faster than the FireWire 400 drive. When duplicating a folder filled with 1GB of files and subfolders, the FireWire 800 drives were 23 percent faster. When copying a file from the internal boot drive to the external FireWire drive, the FireWire 800 drives were 30 percent faster than the FireWire 400 drive. Other tests didn't show as much of a difference. The FireWire 800 drives copied a 1GB folder of files to the internal boot drive 17 percent faster than the FireWire 400 drive. Backing up the same 1GB folder from the internal drive to the FireWire 800 drives with Retrospect was a bit more than 9 percent faster than with a FireWire 400 drive.

For what I'd be doing it looks like an external Oxford bridged Firewire 800 device is really not significantly faster than Firewire 400 device -- even on Mac with Firewire 800 support.

OWC Mercury Elite Pro Firewire 400 + USB 2.0 Combo Kit - FAILS!

Other World Computing- Item Info

[5/05: see also this later discussion.]

Now this might be a contender. OWC is a longstanding Mac vendor with a good reputation. This enclosure has an Oxford 911 chipset -- known to be problematic but at least familiar. They don't bundle Intech but they mention it. The price seems right at $90. It's fanless but they mention heat (at least!) and it has obvious cooling holes. They claim support up to 500 GB (meaning hot drives).

It's one of the few Firewire 400 enclosures sold these days -- so no need to buy a Firewire 400/800 cable or adapter. Includes firewire cable. Also, I have a genuine Apple cable floating around unused.

OWC provides a web page for downloading an Oxford updater.

The assembly directions show real metal in contact with the drive -- so there's HOPE that it won't cook the Maxtor.

All good signs. I'll research further.

Update: I've ordered this drive. They provided an OEM version of the Intech software for $10 (no lifetime updates of course) so I bit on that too. Also a clearance special on a 2.5 drive case, open box, USB only -- for $9. Shipping was $5 for next-day UPS. If this works OWC has made a big fan!

Update: Were I to do this all over again, I'd probably buy a LaCie drive/enclosure unit together. That said, I'm pretty happy so far with OWC. This is not a big operation -- the return address on the plain box had a person's name on the label - presumably the owner, shipper, assembler, etc. The case box was plain white, and completely bare of ANY directions, warrantee, etc. I presume this was a packing error as the web site has quite good assembly directions. In any event this isn't rocket science, Everything was in the box, including extra screws.

Assembly was much more intricate and solid that the CompUSA drive-destructo unit I used last time. Ventilation is interesting and a bit more complex than I'd imagined, it looks like there are a series of plastic baffles to keep dust out. In this case the drive sound is audible in a quiet room. This is a feature -- it means the drive is not completely sealed from the world. The drive/case never gets more than mildly warm. Wow.

With the cheapo cruddy enclosure my large file copy test failed and locked up the system. With this enclosure/bridge set the test completed very quickly and without error. The cheapo enclosure's Prolific PL3507 bridge, and//or the extreme heat, caused delayed and incomplete eject/dismounts. This time the drive promptly dismounted and spun down instantly.

The case is rugged, attractive and very stable with solid and protective bumper/drive stands. It has a slighly eerie bluish glow from the front diode -- an effect wasted on old guys like me.

I'm a happy customer -- even if there weren't any instructions in the box!

Update: This once this looked like a good solution, but it failed the mount, dismount and sleep tests.

In contrast, the LaCie 160GB firewire drive passed all the tests. LaCie does NOT use an Oxford chipset. I think the Oxford chipset does not work properly on many Macs.

Given OWC's 15% restocking fee and the hassles of shipping, I may just use this enclosure with my desktop XP machine. I'm going to buy a LaCie 160GB for my iBook.

Lessons for Firewire and the Mac:
1. It's very problematic -- due to chipset/software issues.
2. Avoid Oxford and avoid Prolific chipsets.
3. Buy an integrated drive/enclosures, don't try to get a random drive to work even with a non-Oxford, non-Prolific chipset.
Update 5/14/05:

The saga continues, now with a new twist. I use removable Vantec drive bay catridges containing 3.5" hard drives for our home LAN backup. I rotate on cartridge off-site. I've been using 80GB Western Digital WD800 ATA drives, but they're no longer large enough. I bought 200 GB Seagate Barracuda drives as replacements. Alas, they didn't work in the Vantec cartridges though they work well in a regular drive bay.

Then I put the 200 GB Barracuda in the OWC Elite Firewire/USB kit. It worked fine. Then I tried the "sleep test". It passed! So ... the scorecard looks like this:
  • OWC firewire enclosure + Maxtor DiamondPlus 9 200 GB: works, but if laptop sleeps drive dismounts (rudely)
  • Vantec enclosure + Seagate Barracuda 200GB: doesn't work at all
  • OWC firewire enclosure + Seagate Barracuda 200GB: works and sleeps
  • Vantec enclosure + Western Digital WD800 80GB: works

Intech SpeedTools Utilites for MacOS X - Problems with Firewire!

Intech Software Online:Product Info:SpeedTools Utilites

After my firewire debacle (see prior posts) I have a renewed interest in this sort of product. This does a lot of different things, including allowing one to create huge FAT32 drives from OS X.

It's not cheap at $90, but it may be bundled with some drives.

It includes the very interesting SpeedTools Device Tuner (for Oxford chipset bridged firewire enclosures)
DeviceTuner was initially created to solve the problems of a particular FireWire device’s ability to transfer data reliably over a FireWire bus. Being unable to copy files to or from a FireWire hard disk, for example, is a common symptom of a larger problem: less than optimal signal quality on the FireWire bus. This can be caused by a myriad of factors, but regardless of why it happens, DeviceTuner can likely help.

FireWire troubleshooting education can quickly become quite complex. Therefore, if you have interest in the "nuts and bolts" of how this product works, Intech strongly recommends that you download the Device Tuner User's guide from the link near the bottom of this page.

...

Controls FireWire payload sizes on an individual or global basis
FireWire "Safe Mode" for maximum compatibility

...
Data is exchanged between FireWire devices in discrete packets of various size. Currently, these packet sizes (called payload) can vary from as little as 4 bytes to as much as 4096 bytes. Most FireWire devices default to anywhere between 1024 to 4096 bytes per packet to achieve maximum speed. This works well for most devices connected to Macintosh computers. However, as more devices get added to a FireWire bus, signal quality can degrade below the threshold that will allow larger data packets to transfer reliably. Furthermore, older Macs (like first generation PCI Power Macs) using older PCI FireWIre host adapters seem to also frequently have marginal signal quality which result in data transfer errors with larger packets.

In situations like these, DeviceTuner can make the difference between data transfer errors and perfect data transfer. Device Tuner allows you to restrict the maximum size of a data packet that a given supported device will use. It works by overriding the payload sizes negotiated between the MacOS and the device. Unfortunately, this tactic does not work with all devices, especially early FireWire disk drives. Note: Be aware that there is always a tradeoff: the lower the packet size, the slower the maximum speed, and visa versa. In a very real sense, Device Tuner does NOT solve the bus's underlying signal quality degradation problem. It just allows you to reduce a device’s FireWire bus payload by reducing packet sizes below the levels which manifest data transfer problems.

This reminds me all too much of my ancient PowerBook 165 SCSI bus experiences. Sigh. One needs the very best cables -- all from one vendor. The very best devices. The very best patience ...

Lessons from the abortive OS X external firewire enclosure debacle

Quick Notes Blog: September 2004

So, what did I learn? Or, in the first two instances, learn again?

1. Don't confuse cost of purchase with cost of ownership.

The $50 enclosure was about $100 too expensive -- and that's only because the probems showed up immediately. In other words, if they paid me $50 to take it it would still be too expensive.

2. Pay in proportion to importance.

The main utility of an external Firewire device would be to hold all my images -- which have swamped my 15GB internal iBook drive. Those images are very valuable. Even with the redundant backup methods I use (optical, off-site, on-site) I shouldn't have risked them with a low-end primary storage device.

3. Firewire is a problem.

Firewire is primarily a Mac solution. It's been around for over five years. There have been problems with every release of OS X. There have never been consumer oriented true firewire devices -- only ATA devices with 1394 bridges. There have been several cycles of firmware updates.

After a while, one should catch on that there's something wrong with this picture -- especially when implemented on a laptop. Maybe Firewire is just not suited to the consumer marketplace. Maybe USB 2 really is the better solution (too bad my iBook is USB 1!).

End of the Line for my cheapo CompUSA OS X drive enclosure: Prolific PL3507 Combo Firewire Device

julian/blog � Prolific PL3507 Firewire Device

I bought a cheapo $50 usb/firewire enclosure at CompUSA, stuck in a very cheap Maxtor 200GB drive and started having problems.

I couldn't identify the chipset on my Mac, but XP did well with the properties tab on the drive manager utility. It's a
Prolific PL3507 Combo Device (1394-ATAPI rev1.10) IEEE 1394 SBP2 Device
The Taiwanese vendor site has been rumored to have firmware updates in the past, but as of today there are none listed.

Searching the net I found several similar loooking enclosures sold under different names, typically from $50-$80.

Other web searches found bad news on every front with both the Maxtor drive and this chipset. The Maxtors don't seem to work that well in Firewire enclosures, and this ATA-Firewire bridge chipset is notoriously problematic under every possible OS. So the enclosure is going back to CompUSA.

When I pulled the drive from the enclosure, the enclosure felt cool to the touch. The drive, however, was too hot to hold. It even smelled slighly burnt. Although the enclosure has an aluminum exterior, the innards are plastic. There is no conductive channel between the drive and the enclosure. The extereme heat alone may have caused the problems I was seeing. I suspect the drive would have died within a few weeks.

Firewire is clearly problematic. I wonder if the technology is fundamentally flawed -- Apple has had problems with these drives under OS X since day one. Maybe that's why the new iMac only supports Firewire 400 and not Firewire 800. Perhaps Apple is resigned to switching to USB 2 or a future alternative.

PS. After scanning the Macintouch Firewire report I'm thinking I don't want to attach anything but the most costly and exotic blessed firewire device with cables from heaven to my lowly iBook.

The firewire bug website for OS X

Original Firewire Bug
So how do I figure out which chipset is IN my cheapo firewire case? Hmm.

Firewire Direct 3.5 inch drive enclosure

Ultra III 800 FireWire 800 USB 2.0 hard drives and enclosures for DV, Digital Audio, Digital Video & Graphic Pros

They bundle MacDrive.