S T 1 5 1 5 0 F C B A R R A . - 4 F C SEAGATE Native| Translation ------+-----+-----+----- Form 3.5"/HH Cylinders 3711| | | Capacity form/unform 4294/ 5062 MB Heads 21| | | Seek time / track 8.5/ 0.8 ms Sector/track | | | Controller FIBRE CHANNEL DUAL Precompensation Cache/Buffer 1024 KB MULTI-SEGMEN Landing Zone Data transfer rate 6.000 MB/S int Bytes/Sector 512 100.000 MB/S ext Recording method RLL 1/7 operating | non-operating -------------+-------------- Supply voltage 5/12 V Temperature *C 5 50 | -40 70 Power: sleep W Humidity % | standby W Altitude km | idle 18.4 W Shock g | seek W Rotation RPM 7200 read/write W Acoustic dBA spin-up W ECC Bit MTBF h 800000 Warranty Month 60 Lift/Lock/Park YES Certificates CSA,EN60950,FCC,IEC950,UL1... ********************************************************************** L A Y O U T ********************************************************************** SEAGATE ST15150FC INSTALLATION GUIDE 83329070, REV. A 9/95 +---------------------------------------------+ | | | | | | | INTERFACE | | +-----------------------+ | +----------+XXXXXXXXXXXXXXXXXXXXXXX+----------+ +-----------------------+ 40-PIN ********************************************************************** J U M P E R S ********************************************************************** SEAGATE ST15150FC INSTALLATION GUIDE 83329070 REV. A 9/95 Jumper Setting ============== REAR VIEW 40 pin I/O and DC Power Wall/bracket Connection | Female +-----|----------+ TOP (HDA) ----------+ ++++++++++++++ +----------- +-1------------20+ BOTTOM Notes on 40 pin I/O connector: +12V = pins 2, 3, 4, 21 +5V = pins 19, 20, 40 GND = pins 6, 22, 23, 26, 29, 32, 35 Mating FC connector: AMP US p/n: 787317-1 straight-in, Male 40-pin FRONT VIEW ---------- Reserved. Shipped with cover installed. Do not remove. Do not install jumpers! | | ++---- RESERVED | ||+--- Fault LED | |||+-- Port B Bypass LED | ||||+- Port A Bypass LED TOP (HDA) ++---+-++++1+ ---PCB----------------------+::::|::::::+-------- ++ BOTTOM +-J6-++++---+ ++ |||+++ | ||| +-- Ground LED + ||+---- Active LED |+----- Ground +------ Remote LED (pin-11 +5v) ST15150FC drives have two ports for connection to two independent loops. Both loops may be active, but only one of these ports may be receiving or originating data at any one time. Do not connect both ports to the same loop. Connecting remote LEDs ---------------------- NOTE THE LYJX-0 BOARD DOES NOT HAVE THE J20 CONNECTOR!! You can connect remote LEDS using J20. Connect the anode (usually the longer LED connector) to the +5V pin, and the cathode to the appropriate LED output pin. For example, if you want to attach an LED which lights up when the drive is active (reading or writing), connect the LEDs anode connector to J20 pin 6 and the cathode to J20 pin 3. +-------------------+------------------------------------------+ | | | | +---+ | | +---+ | | | | | +-----+J20 | | | |* * 1| | | | |* * 2| | | | +-----+ | | +-------------------+------------------------------------------+ +-----+J20 |o o *| Port A Bypass LED |o o o| +-----+ +-----+J20 |o o o| Port B Bypass LED |o o *| +-----+ +-----+J20 |o * o| Active LED |o o o| +-----+ +-----+J20 |o o o| Fault LED |o * o| +-----+ +-----+J20 |* o o| GROUND |o o o| +-----+ +-----+J20 |o o o| +5 VOLT |* o o| +-----+ ********************************************************************** I N S T A L L ********************************************************************** SEAGATE ST15150FC INSTALLATION GUIDE 83329070 REV. A 9/95 Notes On Installation ===================== Drive orientation ----------------- The drive may be mounted in any orientation. All drive performance characterizations, however, have been done with the drive in horizontal (discs level) and vertical (drive on its side) orienta- tions, which are the two preferred mounting orientations. Installation direction ---------------------- horizontally vertically +-----------------+ +--+ +--+ | | | +-----+ +-----+ | | | | | | | | | +-+-----------------+-+ | | | | | | +---------------------+ | | | | | | | | | | | | | | | | | | +---------------------+ | +-----+ +-----+ | +-+-----------------+-+ +--+ +--+ | | | | +-----------------+ The drive will operate in all axis (6 directions). Installation ------------ ST15150FC disc drive installation is a plug-and-play process. There are no jumpers, switches, or terminators on the drive which need to be set. Simply plug the drive into the host's 40-pin Fibre Channel backpanel connector (FC-SCA)-no cables are required. The FC-AL interface is used to select drive ID and all option configurations for devices on the loop. If multiple devices are on the same FC-AL and physical addresses are used, set the device selection IDs (SEL IDs) on the backpanel so that no two devices have the same selection ID. This is called the hard assigned arbitrated loop physical address (AL_PA). There are 125 AL_PAs available. If you set the AL-PA on the backpanel to any value other than 0, the device plugged into the backpanel's SCA connector inherits this AL_PA. In the event you don't success- fully assign unique hard addresses (and therefore have duplicate selection IDs assigned to two or more devices), the FC-AL generates a message indicating this condition. If you set the AL-PA on the backpanel to a value of 0, the system issues a unique soft- assigned physical address automatically. Loop initialization is the process used to verify or obtain an address. The loop initialization process is performed when power is applied to the drive, when a device is added or removed from the FC loop, or when a device times out attempting to win arbitration. - Set all option selections in the connector prior to applying power to the drive. If you change options after applying power to the drive, recycle the drive power to activate the new settings. - It is not necessary to low level format this drive. The drive is shipped from the factory low level formatted in 512-byte sectors. You need to reformat the drive only if you want to select a different sector size or if you select a different spare sector allocation scheme. Do not touch the connector pins or any components on the control board without observing static-discharge precautions. Always handle the drive by the frame only. Mount the drive to the host system's chassis using four 6-32UNC screws. Two mounting holes are in each side of the drive and there are four holes in the bottom of the drive. Cooling ------- Cabinet cooling must be designed by the customer so that the ambient temperature immediately surrounding the drive will not exceed temperature conditions. Air flow -------- The rack, cabinet, or drawer environment for the Barracuda 4FC drive must provide cooling of the electronics and head and disc assembly (HDA). You should confirm that adequate cooling is provided using the temperature measurement guidelines described below. The drive should be oriented, or air flow directed, so that the least amount of air flow resistance is created while providing air flow to the electronics and HDA. Also, the shortest possible path between the air inlet and exit should be chosen to minimize the travel length of air heated by the drive and other heat sources within the rack, cabinet, or drawer environment. The air flow patterns are created by one or more fans, either forcing or drawing air as shown in the illustrations. Other air flow patterns are acceptable as long as the temperature measurement guidelines are met. To confirm that the required cooling for the electronics and HDA is provided, place the drive in its final mechanical configuration, perform random write/read operations and, after the temperatures stabilize, measure the case temperature of the components listed below. Drive mounting -------------- Mount the drive using the bottom or side mounting holes. If you mount the drive using the bottom holes, ensure that you do not physically distort the drive by attempting to mount it on a stiff, non-flat surface. The allowable mounting surface stiffness is 80 lb/in (14.0 N/mm). The following equation and paragraph define the allowable mounting surface stiffness: 80 lb in k = = or F x 14.0 N mm where k is the mounting surface stiffness (units in pounds or newton) and x is the out-of-plane distortion (units in inches or millimeters). The out-of-plane distortion (x) is determined by defining a plane with three of the four mounting points fixed and evaluating the out-of-plane defection of the fourth mounting point when a known force (F) is applied to the fourth point. Grounding --------- Signal ground (PCB) and HDA ground are connected together in the Barracuda 4 family drives, do not separate this connection. Maximizing the conductive contact area between HDA ground and system ground may reduce radiated emissions. A bracket shield with tapped holes is available to system integrators. This shield makes it easier to attach a braid or similar high-frequency grounding device. If you do not want the system chassis to be connected to the HDA/PCB ground, you must provide a nonconductive (electrically isolating) method of mounting the drive in the host equipment; however, this may increase radiated emissions and is the system designer's responsibility. Termination ----------- The reference index signal (SSREF+) is terminated with a 2.21K ohm resistor. Each drive has a termination resistor located on the Main PCB. The terminator resistor is not removable and is always in the circuit. Back-feeding of current is prevented by a diode. Cache operation --------------- Of the 1,024 Kbytes physical buffer space in the drive, 998 Kbytes can be used as a cache. The cache can be divided into logical segments from which data is read and to which data is written. The drive keeps track of the logical block addresses of the data stored in each segment of the cache. If the cache is enabled (see RCD bit in the Fibre Channel Arbitrated Loop Product Manual), data requested by the host with a read command is retrieved from the cache, if possible, before any disc access is initiated. Data in contiguous logical blocks immediately beyond that requested by the Read command can be retrieved and stored in the cache for immediate transfer to the initiator on subsequent read commands. This is referred to as the prefetch operation. Since data that is prefetched may replace data already in the cache segment, an initiator can limit the amount of prefetch data to optimize system performance. The drive never prefetches more sectors than the number specified in bytes 8 and 9 of Mode page 08h (see Fibre Channel Arbitrated Loop Product Manual). If the cache is not enabled, 998 Kbytes of the buffer are used as a circular buffer for read/ writes, with no prefetch operation and no segmented cache operation. The following is a simplified description of the prefetch/cache operation: Case A_read command is received and the first logical block is already in cache: 1. Drive transfers to the initiator the first logical block requested plus all subsequent contiguous logical blocks that are already in the cache. This data may be in multiple segments. 2. When a requested logical block is reached that is not in any segment, the drive fetches it and any remaining requested logical block addresses from the disc and puts them in a segment of the cache. The drive transfers the remaining requested logical blocks from the cache to the initiator in accordance with the "buffer-full" ratio specification given in Mode Select Disconnect/ Reconnect parameters, page 02h (see the Fibre Channel Arbitrated Loop Product Manual). 3. The drive prefetches additional logical blocks contiguous to those transferred in step 2 above and stores them in the segment. The drive stops filling the segment when the maximum prefetch value has been transferred (see the Fibre Channel Arbitrated Loop Product Manual). Synchronized spindle operation ------------------------------ Synchronized spindle operation allows several drives operating from the same host to operate their spindles at the same synchronized rotational rate. Drives operating in a system in synchronized mode increase the system capacity and transfer rate in a cost-effective manner. Each drive in the system can be configured by the host (using a Mode Select command) to operate in either the master or slave mode. Drives can be reconfigured by the host any time after power-up to be master or slave by use of the Mode Select command Rigid Disc Drive Geometry page. The master provides the reference signal to which all other drives phaselock, including the master. There is only one master per system, and that can be a drive or the host computer. All drives may be configured as slaves allowing the host to provide the reference signal. Each drive can be configured for the nonsynchronized mode in which it ignores any reference signal that might be present_this is the default mode as shipped from the factory. The connection of the synchronized reference signal to the host is required only if the host is to provide the reference signal. If the host does not provide the reference signal, the host should not be connected. Hot plugging the drive ---------------------- Inserting and removing the drive on the FC-AL will disrupt loop operation. The disruption occurs when the receiver of the next device in the loop must synchronize to a different input signal. FC error detection mecha-nisms, character sync, running disparity, word sync, and CRC are able to detect any error. Recovery is initiated based on the type of error. The Barracuda 4FC disc drive defaults to the FC-AL Monitoring state, Pass-through state, when it is powered-on by switching the power or hot plugged. The control line to an optional port bypass circuit (external to the drive), defaults to the Enable Bypass state. If the bypass circuit is present, the next device in the loop will continue to receive the output of the previous device to the newly inserted device. If the bypass circuit is not present, loop operation is temporarily disrupted until the next device starts receiving the output from the newly inserted device and regains synchronization to the new input. The Pass-through state is disabled while the disc performs self test of the FC interface. The control line for an external port bypass circuit remains in the Enable Bypass state while self test is running. If the bypass circuit is present, loop operation may continue. If the bypass circuit is not present, loop operation will be halted while the self test of the FC interface runs. When the self test completes successfully, the control line to the bypass circuit is disabled and the drive enters the FC-AL Monitoring state, Pass-though state. The receiver on the next device in the loop must synchronize to output of the newly inserted drive. If the self test fails, the control line to the bypass circuit remains in the Enable Bypass state. Note: It is the responsibility of the systems integrator to assure that no temperature, energy, or voltage hazard is presented during the hot connect/disconnect (hot plug) operation. Discharge the static electricity from the drive carrier prior to inserting it into the system. Temperature ----------- a. Operating The MTBF specification for the drive (800,000 hours) is based on operating at a local ambient temperature of 95*F (35*C). Occasional excursions to drive ambient temperatures to 122*F (50*C) may occur without impact to specified MTBF. The enclosure for the drive should be designed such that the case temperatures at the locations specified in Figures 11 and 12 are not exceeded. Air flow is needed to achieve these temperature values. Continual or sustained operation at case temperatures above these values may degrade MTBF. The drive meets all specifications over a 41*F to 122*F (5*C to 50*C) drive ambient temperature range with a maximum gradient of 36*F (20*C) per hour when the case temperature limits specified above are not exceeded. b. Non-operating Non-operating temperature should remain between -40*F to 158*F (-40*C to 70*C) package ambient with a maximum gradient of 36*F (20*C) per hour. This assumes that the drive is packaged in the shipping container designed by Seagate. ********************************************************************** F E A T U R E S ********************************************************************** SEAGATE ST15150FC INSTALLATION GUIDE 83329070 REV. A 9/95 Barracuda 4FC drives support the Fibre Channel Arbitrated Loop and SCSI Fibre Channel Protocol specifications to the extent described in this manual. The Fibre Channel Arbitrated Loop Product Manual (part number 77767496) describes the general Fibre Channel Arbitrated Loop characteristics of this and other Seagate Fibre Channel drives. Standards --------- The Barracuda 4FC disc drive is designed to be a UL recognized component per UL1950, CSA certified to CSA C22.2 No 950-M89, and VDE certified to VDE 0805 and EN60950. The Barracuda 4FC disc drive is supplied as a component part. It is the responsibility of the subsystem designer to meet EMC/ regulatory requirements established by the FCC, DOC, and VDE. Engineering test characterizations of radiated emissions are available from the Seagate safety department. General description ------------------- Barracuda 4FC drives are random access storage devices designed to support the Fibre Channel Arbitrated Loop (FC-AL) and SCSI Fibre Channel Protocol as described in the ANSI specifications, this document, and the Fibre Channel Arbitrated Loop Product Manual (part number 77767496) which describes the general interface characteristics of this drive. You can view the Fibre Channel interface simply as a transport vehicle for the supported command set (ST15150FC drives use the SCSI command set). In fact, the Fibre Channel interface is unaware of the content or meaning of the information being transported. It simply packs the SCSI commands in packets, transports them to the appropriate devices, and provides error checking to ensure that the information reaches its destination accurately. The head and disc assembly (HDA) is environmentally sealed at the factory. Air recirculates within the HDA through a non-replaceable filter to maintain a contamination-free HDA environment. Never disassemble the HDA. This exploded view is for information only. Do not attempt to service items in the sealed enclosure (heads, media, actuator, etc.) as this requires special facilities. The drive contains no parts replaceable by the user and opening the HDA for any reason voids your warranty. Barracuda 4FC drives use a dedicated landing zone at the innermost radius of the media to eliminate the possibility of destroying or degrading data by landing in the data zone. The heads automatically go to the landing zone when power is removed from the drive. An automatic shipping lock prevents potential damage to the heads and discs that results from movement during shipping and handling. The shipping lock disengages and the head load process begins when power is applied to the drive. Barracuda 4FC drives decode track 0 location data from the dedicated servo surface to eliminate mechanical transducer adjustments and related reliability concerns. The drives also use a high-performance actuator assembly design that provides excellent performance with minimum power dissipation. Standard features ----------------- Barracuda 4FC drives have the following standard features: - Integrated dual port FC-AL controller - Support for FC-AL (Fibre Channel Arbitrated Loop) - Differential copper FC drivers and receivers - Downloadable firmware using the FC-AL interface - Drive selection ID and configuration options are set on the FC-AL backpanel, T-card, or through interface commands. Jumpers are not required on the drive. - FC world-wide name uniquely identifies the drive and each port - Supports up to 16 initiators - User-selectable logical block size (180 to 4,096 bytes) - Reallocation of defects on command (Post Format) - User-selectable number of spare sectors per cylinder - Industry standard 3.5-inch full-high form factor dimensions - Programmable sector reallocation scheme - Flawed sector reallocation at format time - Programmable autowrite and read reallocation - Reallocation of defects on command (post format) - 96-bit Reed-Solomon error correction code - Sealed head and disc assembly (HDA) - No preventive maintenance or adjustments required - Dedicated head landing zone - Automatic shipping lock - Automatic thermal compensation - Embedded Grey Code track address to eliminate seek errors - Self-diagnostics performed at power on - 1:1 interleave - Zone bit recording (ZBR) - Vertical, horizontal, or top down mounting - Dynamic spindle brake - 998 Kbyte data buffer Media description ----------------- The media used on the drive has a diameter of approximately 95 mm (approximately 3.7 inches). The aluminum substrate is coated with a thin film magnetic material, overcoated with a proprietary protective layer for improved durability and environmental protection. Performance ----------- - Programmable multi-segmentable cache buffer - 106.3 Mbytes/sec maximum instantaneous data transfers - 7,200 RPM spindle; average latency = 4.17 msec - Command queuing of up to 64 commands - Background processing of queue - Supports start and stop commands - Provides synchronized spindle capability - Adaptive seek velocity; improved seek performance Reliability ----------- - 800,000 hour MTBF (Class A computer room environment) - Fibre Channel (FC) interface transports SCSI protocol through CRC protected frames - LSI circuitry - Balanced low mass rotary voice coil actuator Unformatted and formatted capacities ------------------------------------ The standard OEM models are formatted to 512 bytes per block. ST15150FC drives have nine (9) spare sectors per cylinder and one spare cylinder per unit. Users having the necessary equipment may modify the data block size before issuing a format command and obtain different formatted capacities than those listed. User-available capacity also depends on the spare reallocation scheme you select. See the Mode Select command and the Format command in the Fibre Channel Arbitrated Loop Product Manual (part number 77767496). Factory-installed accessories ----------------------------- OEM standard drives are shipped with the Barracuda 4FC Installation Guide (part number 83329070). Factory-installed options ------------------------- You may order the following items which are incorporated at the manufacturing facility during production or packaged before shipping: - Black plastic front panel with green lens (part number 70553702).* - Black plastic front panel with red lens (part number 70553701).* - Single-unit shipping pack. The drive is normally shipped in bulk packaging to provide maximum protection against transit damage. Units shipped individually require additional protection as provided by the single unit shipping pack. Users planning single unit distribution should specify this option. * You may order other front panel colors. Each panel has a single rectangular LED indicator lens that, when glowing, indicates the drive is selected. User-installed accessories -------------------------- The following accessories are available. All kits may be installed in the field. - Front panel kit (green lens), part number 70869751. - Single-unit shipping pack kit. - Adapter accessory frame kit, part number 75790701. (adapts a 3.5-inch drive to fit in a 5.25-inch drive mounting space). This kit contains the frame to allow a 3.5-inch drive to be mounted in a 5.25-inch drive bay. It includes mounting hardware, front panel with a green lens, an LED with cable that connects to the remote LED connector, and installation instructions. - Evaluation kit, part number 70935895. This kit provides an adapter card ("T-card") to allow cable connections for two FC interfaces and DC power. Two twin axial cables, 6-feet in length, are included for the input and output connections to the FC interfaces. A small DC fan is included for cooling. All performance characteristics assume that thermal calibration is not in process when the SCSI command is received. A SCSI command being executed is not interrupted for thermal calibration. If thermal calibration is in process when a SCSI command is received, the command is queued until the compensation for the specific head being calibrated completes. When compensation completes for the specific head being calibrated, the first queued SCSI command is executed. Thermal calibration ------------------- ST15150FC drives use an automatic thermal calibration (TCAL) process to maintain accurate head alignment with the data cylinders. The host system may choose to allow the drive to perform TCAL at the drive's predefined intervals or the Rezero Unit command may be issued by the host to reset the TCAL timer so that the host knows when the TCAL will occur. 1. At power up and following a SCSI reset, the drive calibrates all of the heads before any read or write com-mands are processed. All heads are also calibrated during the SCSI Rezero Unit command. 2. The drive delays 300 seconds before initiating any TCALs. No TCALs occur during this delay period. 3. A single-head TCAL is then scheduled at 7.1 second intervals. 4. After the drive TCALs all of the heads, the interval is increased to schedule a single head TCAL every 14.3 seconds. 5. The drive attempts to find an idle period of 25 to 50 milliseconds prior to performing a single head TCAL. If this TCAL is delayed for another interval of time, the drive forces the TCAL at the next command boundary. This guarantees that no head will remain uncalibrated for more than 600 seconds (2 * 21 heads * 14.3 seconds per head) and that no TCALs are closer together than the interval time. Note. Any TCAL performed during the "standard" retry sequence is limited to the failing head and is disabled if the host has selects a retry count of zero. Defect and error management --------------------------- The drive, as delivered, complies with this product manual. The read error rates and specified storage capaci-ties are not dependent upon use of defect management routines by the host (initiator). Defect and error management in the SCSI protocol involves the drive internal defect/error management and FC-AL system error considera- tions (errors in communications between the initiator and the drive). Tools for use in designing a defect/error management plan are briefly outlined in this section. References to other sections are provided when necessary. Drive internal defects/errors ----------------------------- Identified defects are recorded on the drive defects list tracks (referred to as the primary or ETF defect list). These known defects are reallocated during the initial drive format operation at the factory. See the Format Unit command in the Fibre Channel Arbitrated Loop Product Manual (part number 77767496). Data correction by ECC is applied to recover data from additional flaws if they occur. Details of the SCSI commands supported by the drive are described in the Fibre Channel Arbitrated Loop Product Manual. Also, more information on the drive Error Recovery philosophy is presented in the Fibre Channel Arbitrated Loop Product Manual. Physical description -------------------- ST15150FC drives may be connected in a loop together or with other compatible FC-AL devices. A maximum of 127 devices may have addresses; however, one of the addresses is reserved for a fabric port switch device. This means 126 addresses are available for FC-AL devices. More FC-AL compatible devices may physically reside on the loop, but they will not be functional because they would not be able to obtain valid addresses. Port bypass circuits (PBCs) allow devices to be inserted into unpopulated locations or removed from the loop with loop opera- tion recovery after a brief disruption. These PBCs are located external to the FC-AL device. Power ----- Power is supplied through the FC-SCA with support for +5 volts and +12 volts. All of the voltage pins in the drive connector are the same length. Four 12 volt pins provide +12 volt power to the drive. The current return for the +12 volt power supply is through the common ground pins. The supply current and return current must be distributed as evenly as possible among the pins. The maximum current typically occurs while the drive motor is starting. Three 5 volt pins provide logic power to the drive. The current return for the +5 volt power supply is through the common ground pins. The supply and return current must be distributed as evenly as possible among the voltage and ground pins. The mating connector pins use shorter contacts to achieve power surge reductions and to aid in "hot plugging" the drives. There are longer voltage contacts in the connector to enable the drive filter capacitors to charge. Current to the drive through the long charge pins is limited by the system in which the drive operates. Three of the +12 volt pins are shorter to allow capacitive pre- charging through the longer +12 volt charge pin. Two of the +5 volt pins are shorter to allow capacitive precharging through the longer +5 volt charge pin. Fault LED out ------------- The Fault LED Out signal is driven by the drive when: - the drive detects failure of both ports - the drive detects an internal disc failure - the drive receives the appropriate fault LED command from the host The Fault LED Out signal is designed to pull down the cathode of an LED. The anode is attached to the proper +5 voltage supply through an appropriate current limiting resistor. The LED and the current limiting resistor are external to the drive. Synchronized spindles interface ------------------------------- The synchronized spindles interface (SSI) allows several drives operating from the same host to operate their spindles at a synchronized rotational rate. Electrical description of the SSI --------------------------------- The electrical interface consists of one digital TTL reference index signal and ground. The reference index signal (SSREF+) is an output if the drive is configured as a master and is an input otherwise. ********************************************************************** G E N E R A L ********************************************************************** SEAGATE FC-AL INTERFACE An Overview of Fibre Channel --------------------------- Introduction ------------ Everyone has accepted the fact that we have moved into the Age of Information. In this paradigm information itself is a commodity, and therefore there is great value in its efficient disbursement. Unfortunately, industry has placed greater value in creating information, than distributing it. We often hear about new machines which are capable of performing prodigious calculation at the blink of an eye. New reports of ever faster computers are commonplace. Sharing this information, however, has become a priority only recently. It seems that although we have moved into the Age of Information, one of our biggest challenges is to efficiently distribute the information for everyone to use. Luckily, a viable solution is at hand. Conceived and supported by such industry giants as IBM, Hewlett-Packard, and Sun Microsystems, the Fibre Channel is aimed at providing an inexpensive, flexible and very high-speed communications system. Most of the popular network implementations today can claim to have any two of these elements. Since Fibre Channel encompasses all three, it has everything necessary to become a resounding success. Not the Network Fibre Channel has significant advantages over common networks. The first difference is speed. The fastest network implementations today support transfer data at a little over 100 megabits per second. For smaller data files, where a single computer is directly communicating with a file server, such speeds are adequate. However, for realtime video and sound, or systems where two machines must operate on common data even 200 megabits per second is hopelessly inadequate. Fiber Channel provides significantly higher rates, from 10 to 250 times faster than a typical Local Area Network (LAN). In fact, Fibre Channel can transfer data at speeds exceeding 100 megabytes, or 800 megabits, per second. This speed is sufficient to allow transfer of a 1024x768 image with 24-bit color at 30 frame per second, and CD- quality digital sound. This overcomes the bandwidth limitation, which is probably the most serious impediment for LAN performance. As the number of computers communicating on a common network increases, the amount of data packets increases accordingly. This is because data on a LAN is common to all computers on that network. The software must decide if a particular message is relevant for a particular machine. When several machines are communicating with one another, every other machine on the network must contend with all of the messages. As the number of messages increases, the load for the entire system is increased. Fiber channel is a switched system. Much like a telephone system, a connection is established between only the parties that need to communicate. These parties can share the entire bandwidth of Fibre Channel, since they do not have to contend with messages not relevant to their communication. LANs attempt to compensate for this by increasing the transfer speed, which places an even greater burden on the software. Since all protocol for Fibre Channel is handled by the hardware, the software overhead is minimal. Fibre Channel also supports full parallelism, so if greater capacity is needed, more lines can be added. The common analogy for showing the advantages of parallelism is the effect of doubling the number of lanes on a freeway instead of doubling the speed limit. The physical distance between computers is another limiting factor for conventional LANs. Ethernet cables usually have a limit of 1000 feet between machines whereas Fibre Channel can support a link between two up to 10 kilometers apart. Finally, Fibre Channel is not software intensive. All of the essential functions are handled by hardware, freeing the computer's processor to attend to the application at hand. Even the error correction for transmitted data is handled by the Fibre Channel hardware. In standard LANs this requires precious processor resources. Advantages for Computing ------------------------ The obvious advantage for Fibre Channel is to facilitate communication between machines. Several workstations clustered together already surpass the speed and capacity of a VAX, and begin to rival the power of a super computer, at a much lower cost. The power of concurrent processing is awesome. For example, a single neuron inside our brain is much less complex, and operates far slower than a common 286 processor. However, millions of neurons working in parallel can process information much faster than any processor known today. Networking simple logical units, and operating them in parallel offers advantages simply unavailable for the fastest single processor architectures. These shared architectures require a huge amount of communication and data sharing which can only be handled by high-speed networks. Fibre Channel not only meets these requirements, but meets them inexpensively. The hardware industry is partly responsible for the I/O bottleneck. By using the processor speed as the primary focus for their sales efforts, the bus speeds have languished. With respect to the new class of processors, current system bus speeds are greatly lagging. This is something like building a mill which can process 1000 pounds of grain a day, and supplying that mill with a single donkey. There is little use for a fast processor that spends most of its time waiting for data to act upon. Whether this data comes from disc drives, peripherals, or even other processors, today's bus speeds would leave most processors idle, and the next generation of processors will be many times faster. Fiber Channel provides the data transfer capability which can keep current and upcoming processors busy. Impact on Mass Storage ---------------------- Today's fastest interfaces are capable of transferring data at around 20 megabytes per second. However, this speed rating is only for transferring data. All protocol intercommunication occurs at much slower speeds, resulting in a lower effective data transfer rates, typically around 11 megabytes per second. This represents about one-tenth of Fibre Channel's current capability. Fibre Channel drives do not suffer from device protocols occurring at slower speeds, since all communication occurs at 100 megabytes per second, including device intercommunication. In addition to this, the drive itself can be placed up to 10 kilometers away from the computer. This would have two effects on the way mass storage is implemented. First, the amount of data a machine could receive would only be limited to the transfer speed of the drive. For high performance disc arrays this could exceed 50 megabytes per second. Machine and disc storage could finally work to provide real-time, full motion video and sound for several machines simultaneously. With Fibre Channel's ability to work across long distances, these machines could conceivably reside many miles apart. For medical applications, computer design centers, and real-time networks such as reservations systems, this capability would be invaluable. Second, such support for transmitting data over large distances would allow disc drives to be placed away from the computer itself. This would allow for centralized data resource areas within a business office, simplifying everything from site planning to maintenance procedures. Indeed a centralized data resource center would be possible for an entire office complex. The development of the Loop will also provide a huge advantage in implementing large capacity disc sub systems. The Fast/Wide SCSI specification has a theoretical upper limit of 16 total devices attached to a single host. The practical maximum is 6 devices. Fibre Channel supports a theoretical limit of 256 devices for a common host, with a practical implementation of 64 devices. This practical limit is a very conservative figure, and implementation with more devices are easily possible. The Loop allows system designers to build high capacity configurations, well into the terabyte range, with much lower overall cost. Finally, Fibre Channel is a serial communications device which has two immediate advantages. First, the cabling necessary to interconnect Fibre Channel devices is very inexpensive when compared to SCSI cabling. Fibre Channel cabling is also much easier to connect, and replace than SCSI cables, which simplifies the entire process of integration and maintenance for a high capacity data storage system. For corporations that are currently grappling with a the complexity of installation, and high-cost of SCSI cables, this feature will prove invaluable for cutting costs and simplifying installation and upkeep. Secondly, implementing Fibre Channel requires less space on the circuit board than SCSI drives. This reduced space requirement would allow the drive designers to include extended features which cannot currently be implemented. For example, a 3.5-inch form-factor drive with Fibre Channel could be designed with dual-port capability, a feature necessary for use with many mainframes and mini-computers. The space saved on the circuit board by using Fibre Channel would allow for the extra connector and additional circuitry needed for dual-port drives. Conclusion ---------- The Fibre Channel will provide the corporations with data in much the same way the freeway system provided motorists mobility. Access to a vast, interconnected information network which is fast, inexpensive, and flexible. With the adoption of Fibre Channel as an open ANSI standard, its effect on the horizon of computing will be nothing short of revolutionary. We have become very good at processing data; Fibre Channel allows us to move it. The ability to share information will provide the impetus for communication, design and development on a scale not previously possible. By facilitating the fabled data-highway, Fibre Channel will accelerate to the Age of Information, as the steam engine moved us into the Age of Industry.