Serial bus gets serious with USB 2.0: FireWire says: what? Me worry?
Okay, so maybe it isn't the Winter Olympics. But the long-awaited, much-anticipated arrival of USB version 2.0 is upon us, and some observers are already talking about the death of FireWire (IEEE 1394) as the interface of choice for bandwidth-hungry peripherals. But while the throughput improvement for USB is substantial, FireWire is not exactly standing still. And the USB-IF has another new technology in development: this time, Bluetooth may be in the cross hairs.
USB: Ultra Slow Bandwidth?
USB 2.0 is the first major overhaul of the Universal Serial Bus specification since it was introduced seven years ago. At that time, the interface was tasked primarily with supplanting the decade-old serial port, as well as ports like the PS/2 which are primarily used for pointing devices. It was thought that, with 12Mbps throughput, USB would offer plenty of speed for add-ons such as speakers, mice, joy-sticks, CD-ROM drives, and printers. The SCSI-like ability to connect multiple devices to a smart interface--using a USB hub--was revolutionary at a time when most PCs were equipped with just two or three serial ports and a parallel port.
Further, the logic went, the PC would never make it as a home appliance aimed at the non-technical user until the dreaded "case-crack" (the opening of the system to add cards) was eliminated. Sure, the power user and the professional would still tinker with the innards of a system, but USB would enable a whole range of new devices to be purchased and enjoyed by even the casual computer user.
But like most good technologies, USB 1.0 became a victim of its own success. While it took several years before the interface was added to the OS and motherboards, once it was, a landslide of peripherals hit the market: scanners, printers, digital cameras, floppy replacement drives, and PDAs all could now be attached in seconds, with nary a manual in sight. But soon, as users began to transfer huge image and audio files, 12Mbps was just too slow (ever try transferring a few hundred megabytes across a USB connection?)
USB 2.0 is an enormous speed increase over the earlier version of the specification. 2.0 offers throughput of as much as 480Mbps, or 40 times faster than USB 1.1. This means that USB is now appropriate for demanding peripherals such as CD-R, CD-RW, and DVD-R drives, as well as add-on hard drives.
Late last year vendors began shipping USB 2.0-compliant devices. However, tuned USB 2.0 drivers (and systems) have just begun to hit the market, and Microsoft added OS support for Windows XP via a Windows Update download in late January. (At press time, support of USB 2.0 in Windows 2000 was scheduled to be available by late March. Windows 9.x and NT 4.0 will not support version 2.0.)
While some PC OEMs have begun adding USB 2.0 support to systems, the faster interface is expected to truly take off when Intel adds integrated USB 2.0 support to its next core logic chipset, expected by mid-year. Intel has already begun to add 2.0 support to system boards. At CES in January, the company announced two P4 desktop boards (D850MVSE and D845BGSE) that each support up to five USB 2.0 ports.
Lighting The Fire Under FireWire
While version 2.0 of USB was developed to increase the versatility of the original interface, as usual there was also a solid business reason to speed things up. Apple Computer Corp. owns many of the 1394 patents: the interface was developed by Apple engineers in the 1980s and only later submitted to the IEEE for adoption as a standard. The term "FireWire" is, in fact, trademarked by Apple and the company charges a licensing fee of 25 cents for each implementation of the technology. (1394 technology is also known as i-Link, which is trademarked by Sony.) And while USB 2.0 is slightly faster than the original 1394 specification (480Mbps and 400Mbps, respectively), IEEE 1394b (as the newest version is known) will offer a blistering 3.2Gbps of throughput over fiber cable, which puts even USB 2.0 to shame.
"Without USB 2.0, I think you would have seen the printer and camera guys switch to FireWire," says Michael Johas Teener, who was the chief architect of Apple Computer's FireWire development effort in the 1980s. Teener is the originator of the 1394b gigabit effort and is currently Chief Technology Officer at Zayante Inc., a maker of 1394 silicon and software tools.
Unlike USB 2.0, Windows XP supports 1394b out of the box, so no point upgrades are necessary to enjoy it, It should be noted, however, that 1394b-based devices are not expected to hit 3.2Gbps for at least 18 months. Texas Instruments and Agere have demonstrated 1394b-based devices running at 800Mbps; realistically, we can expect to see commercially available products operating at about 800Mbps over copper this year, with a doubling to 1.6Gbps in early 2003 and then another doubling sometime later that year.
1394b also has other advantages over USB: it supports longer distances (100 meters on UTP-5 fiber cable versus about 15 feet for USB without bridging) and is generally considered a more robust interface from a data management perspective. Among its other strengths, 1394 is more suitable for video editing because it provides the ability to mix real-time and asynchronous data on a single connection. However, 1394h has some technical issues that do not plague USB 2.0. "1394b requires new cables and connectors to achieve its higher speeds," says Jason Ziller, Chairman of the USB-Implementers Forum and a technology initiatives manager at Intel. "We do not expect to see PC-based peripherals that require speeds higher than 480Mbps in the next 12-18 months."
Still, no matter what you read in the trade press, USB 2.0 and 1394b are not playing in a zero-sum game. There is more than enough room in the market for both interfaces to thrive; Microsoft acknowledged as much when it decided to support both in Windows XP. (At Comdex, OrangeMicro demonstrated a USB 2/FireWire combination PC Card for notebooks.)
"We will continue to see 1394 and USB 2.0 pursue separate tracks," says Brian O'Rourke, Senior Analyst for Converging Markets & Technologies at Cahners In-Stat. "1394 will be seen primarily in consumer electronics, and USB 2.0 will primarily be seen in PCs and PC peripherals. The strength of 1394 continues to be its ability to stream audio and video in a peer to peer network, while the strength of USB is in its ubiquity, and low price, in the PC world."
Because USB is somewhat more established and generally more widely supported in the PC-oriented device market, it will continue to dominate its more specialized rival in this space. But as long as there is consumer demand, OEMs will continue to offer at least some of their product lines with 1394 interfaces aimed at the video enthusiast market. "FireWire penetration will still occur, it's just going to be slower: these devices still need connectivity to consumer electronics, and these [are dominated by] FireWire," says Zayante's Teener. Intel's Ziller agrees: "We believe that 1394 makes sense for consumer electronics, which is why we encourage it on consumer PCs. But an IS manager needs USB, not a PC with 1394."
In fact, the question is not whether USB will supplant 1394 but rather, will a new serial specification, USB On-The-Go (OTG), cannibalize the nascent Bluetooth market? And here, the answer is far from certain.
USB OTG, a supplement to the 2.0 specification, was formally released by the USB Implementers Forum just before CES. It provides for a new class of USB-compliant device that supports device-to-device connectivity (sometimes called "peering" or "ad hoc networking"). The OTG supplement describes the following enhancements to a typical USB device: limited host capability to communicate with other USB devices without the need for PC intervention; a small USB form factor for highly mobile devices; and low power operation to preserve battery life.
A device with such features is known as a "dual-role" device and can act as either a host or a peripheral when connected to other devices. Unlike a USB 2.0-compliant PC, a dual-role OTG device need not have the capability to identify any other USB device. However, any such dual-role device must be able to operate as a standard compliant peripheral when connected to a PC. (In fact, OTG is not a true peering interface: there is still a master-slave relationship between devices, says the USB-IF's Ziller, so one will still need the necessary logic to fulfill the host role.)
What types of OTG/dual-role devices might we see? The Implementers Forum envisions several classes of device: an MP3 player with the ability to network to other players to share songs; a digital camera that could connect directly to a printer to print photos; or a PDA that might connect to a mobile phone and surf the Web. Ziller says that manufacturers will need to specify what other devices a USB-OTG device will communicate with. Such devices will connect to one another using a specialized cable with a mini connector at both ends, or else will require a dongle at one end of a standard USB cable, according to Ziller.
Bluetooth or Bust?
OTG is seeking to fill the peering role similarly targeted by Bluetooth, which has been slow to market but is finally taking off. Both specifications are designed to transfer data between and among devices, and while both can be used with a PC, neither requires the presence of one to operate. Both technologies are for use in short-haul environments, where devices are within several meters of one another, although in theory (and with no obstructions) Bluetooth devices can exchange data across tens of meters.
This, however, is where the similarities end. Bluetooth is a set of wireless protocols, while USB OTG is a wireline specification. Further, throughput across a Bluetooth connection maxes out at about 800Kbps, while USB is obviously orders-of-magnitude faster. And, of course, OTG-based USB devices are still being developed while Bluetooth products are on store shelves today. "USB OTG is still wired technology, says Joyce Putscher, Director of Converging Markets & Technologies and principal Bluetooth analyst at In-Stat, "[and it still offers] only 5 meters in length."
The rub: Bluetooth has been so delayed that its slow data rate is now only appropriate for a few specialized tasks: phone-to-Internet connectivity and small file transfer, basically. Further, Bluetooth--which operates in the 2.4GHz band--is subject to interference in that crowded portion of the spectrum and has been eclipsed in speed, range, and frequency allocation by 802.11a. Granted, WiFi and other 802.11-based technologies are not designed for ad hoc connectivity, as Bluetooth was. But USB OTG is, and its throughput allows more versatility. "Bluetooth is really just a control protocol that can handle some voice audio," notes Zayante's Michael Teener.
The USB-IF's Ziller does not foresee OTG supplanting Bluetooth but rather envisions portable devices that support both technologies. "Bluetooth, as a wireless technology, is still more convenient [than USB]," he notes.
Will users prefer a tethered connection with 480Mbps throughput to a wireless connection with 1Mbps (which, keep in mind, requires that the devices be in close proximity anyway)? In effect, would they rather have speed or convenience? Or will they end up getting both, depending on the device? It's too soon to know. But it is important to remember that manufacturers have years of experience adding USB support to a wide range of devices (MP3 players, digital cameras, and so on) while Bluetooth chip sets are new to the market. Expect to see OTG silicon sampling this half, with enabled products later this year.
|Printer friendly Cite/link Email Feedback|
|Publication:||Computer Technology Review|
|Date:||Mar 1, 2002|
|Previous Article:||Uncovering the total cost of ownership of storage management.|
|Next Article:||"Frankly, my dear, I don't give a DAM!" Yes, Rhett would, for Digital Asset Management.|