Serial ATA Will Implement Next Generation Performance By 2002--Transparently.

In the beginning, there was MFM and it was good[ldots]for a while. In the early days of smaller-than-Chevrolet sized hard drives, adding a hard drive to a computer was no simple task. Many readers will recall that the original IBM PC was equipped with a connector for a cassette tape drive--the tape drive and a floppy drive were available storage for this first PC.

It took a while for a hard drive to become available--a full height unit that stored a whopping 10MB (that's megabytes) of data. The PC with hard drive was the PCXT. In 1984, the PC AT (Advanced Technology) was announced with a 20MB hard drive standard.

These 20MB drives proved to be tremendously unreliable and, in some ways, threatened to sink the AT. Instead, IBM found a different drive supplier and, instead of the drives sinking the AT, the drives were themselves sunk, dropped into the ocean to eventually provide habitat for aquatic plants and animals.

In those early days, a separate controller card was required in order to connect the drive to the computer. These controllers used an encoding method remembered as MFM. A second controller technology, RLL (Run Length Limited) provided higher capacities--a drive that would provide 20MB of capacity as an MFM drive can store 30MB if configured for RLL.

The high cost of the controller and inherent difficulty connecting a hard drive to a PC were a problem for IBM and most "clone" makers. Tandon Computer developed an embedded controller that made attaching a hard drive to a computer less of a problem. With a standardized controller (which, for a while, was proprietary to Tandon) the difficulty and cost of matching drives to systems improved.

After Western Digital acquired the drive business and interface technologies from Tandon, things became easier for the industry. The basic idea was for the drive manufacturers to put the interface electronics onto the drive and for the system and motherboard makers to put a very simple (and inexpensive) and standard connector onto the motherboard. In effect, the easy and cheap stuff was done on the motherboard, while the more difficult tuning to the hard drives was the responsibility of the drive maker. A standard, IDE (Intelligent Device Electronics) was born. Although some industry folks may try to claim more responsibility for the genesis of IDE than they may actually deserve, the real genesis began at Tandon.

Over the years, IDE and its successors of various versions of ATA, became basic components of most computers. A wide range of devices has been developed for connection via ATA because of the generic nature of the interface. These include CD-ROM drives, CD-Recorders, DVD-ROM and DVD-RAM drives, LS-120 floppy drives and, of course, hard drives.

Over the years, performance of the ATA bus has been stretched--from 33MB/sec to 66MB/sec with 100MB/sec looming close over the horizon. The actual benefits aren't arithmetic, however--a drive with an ATA66 interface won't perform twice as quickly as an ATA33 drive and an ATA100 drive won't be three times as fast as an ATA33. However, relative to each other, drives with a faster interface will transfer data somewhat faster than one with a slower interface--all other things being equal.

The ATA interface, designed to be cheap, has some serious problems. For users who purchase complete computers and don't want to do little more than connect monitor, keyboard, mouse and speakers, and plug the box into a power strip, ATA poses few problems.

However, for those who want to add or replace drives, ATA can be quite challenging. Each ATA connector only supports two drives--a master or a slave. Each device must be set as one or the other. The current ATA design only allows one device to be communicating through the bus at a time. Many users may have learned the hard way that it may not be possible to copy the contents of a CD or CD-ROM onto a hard drive on the same ATA channel because the hard drive may time out, waiting for the CDROM drive to finish reading. The basic answer to this problem is to put the hard drive onto one of two ATA ports and the other device on the second. Another potential problem is that a total of four devices can be attached to a computer. In these days when a user with a CD-ROM drive and one hard drive may want to add a CD-Recorder, a second hard drive, perhaps a DVD-ROM drive and, maybe, even an LS-120 drive, the four device limit may also be a problem.

Additionally, cabling can be quite a headache. Compared to what came before it, the 40 pin ATA connector was an improvement. With the advent of ATA66, the cable connectors now still feature 40 pins with each pin separately grounded (for a total of 80 pins per connector), to provide better isolation against noise. ATA cables are bulky and increasingly expensive. Perhaps even worse, the cables are difficult to route through a computer's case and can make adequate air flow for system cooling a major problem.

Further, the current ATA technology is based on five volt electronics, making ATA a poor candidate for use in notebook computers. Clearly, the move should be towards 3.3 volts or below.

ATA definitely has its strengths. These are primarily ubiquity with their inclusion on a vast majority of motherboards already equipped with ATA in one flavor or another. Low cost to the system makers is another major advantage. Broad software compatibility is a third advantage--most operating systems and most software fully supports ATA devices. Any replacement of ATA should provide similar benefits--low cost, easy implementation, and full software compatibility.

A consortium that includes such major players as Intel, Dell, IBM, Maxtor, Quantum, Seagate, Western Digital, Philips, and a few others are developing this next generation ATA, producing an interface that it currently calls Serial ATA. The data is transmitted serially over the cabling, requiring only two pairs of wires, because it is serial.

Serial ATA will simplify the wiring requirements by physically eliminating the big bulky cable, replacing it with a simple four-wire cable. Interface electronics will manage the flow of data between system and drive.

Serial ATA will connect devices in a star topology with each device using its own connection on the motherboard or host bus adapter card. Initially, only four connections are expected on the motherboard, although there is nothing in the specification that creates a four device limit. Conceivably, some low-end systems may have as few as one or two connections and higher end systems (possibly with ATA RAID) may contain many more than four connections.

The initial transfer rate will be 1.5Gbps--many times that of the current 33Mbps of ATA33 or 66Mbps of ATA66 and fifteen times as fast as ATA100, an 100Mbps upgrade of the parallel ATA specification. The organization anticipates a 2x and 4x version of Serial ATA sometime in the future.

The consortium expects to have a draft specification for Serial ATA later this year. Adoption of the Serial ATA spec is expected to take a year or more with Serial ATA implemented on motherboards and storage devices by 2002.

Additionally, the integration of power connections into the Serial ATA connector, possibly as an added set of wires, is also expected. The ultimate results of the move to Serial ATA and integrated power will be better air flow inside computers; less real estate required for routing ATA cables (enabling even smaller computers) and easier installation and manufacture of devices using Serial ATA. Cable costs should also decline because the cables used will be much less complex than the 40 pin cable used by ATA33 and the 40 pin/80 wire cable used for ATA66. Further, the frustrations experienced by some users who are installing new devices, but use cheap unkeyed connectors, making proper connection a 50% probability, and may have to deal with bent or damaged pins on the motherboard or device, can be avoided by the much simpler, idiot-proof Serial ATA connectors.

Additionally, a single interface can be used for desktop, as well as notebook and handheld devices. Conceivably, small storage devices (like notebook hard drives) will become standard for use in all devices, both portables and desktops. A move from 3.5-inch to 2.5-inch form factors for hard drives has been expected for a number of years--with the adoption of Serial ATA, this change to a new standard form factor may become more practical across the board.

It will take a number of years until Serial ATA replaces Parallel ATA. In the interim period, converters/adapters will be available. The converters will be used as bridges between Serial and Parallel ATA and might appear in a number of places.

For example, a converter may be attached to a legacy motherboard, allowing the motherboard's parallel interface to attach to serial ATA devices. A motherboard with serial connections may use a converter to connect to older parallel drives. Different combinations will be used to enable the mixing of devices and device interfaces within a system.

From a software and operating system standpoint, there should be no impact. The change from Parallel to Serial will be OS transparent--system calls to ATA devices will be handled the same way by either interface.

In the beginning, there was MFM. It was good.

Tomorrow, there will be Serial ATA. It will be much better.