Buffered Recording

U.S. Pat. No. 3,932,894 shows a rotating head recorder and record member with which the present invention can be advantageously employed.

The present invention relates to data recorders and, more particularly, to those data recorders employing cyclic recording and readback operations which may include a portion of a cycle during which no signals are sensed.

As many new products reach the marketplace in the digital signal recording area, there is a continuing desire to increase the areal recording density. While in many instances the record media can retain such increased densities, the data recorders have to be substantially changed. In many instances, there is no forward/backward compatibility between the original low density recording and the new high density recording. Such incompatibilities, while in many instances are unavoidable, add to the cost of signal recording and are to be avoided if at all possible.

Forward/backward compatibility was achieved in one-half inch tape digital data signal records originally when one-half inch tapes used NRZI recording at 200, 556 and 800 bytes per lineal inch. Operator selection was required for the digital data recorders to operate with different density tapes. When phase encoded recording came into being, all phase encoded tapes were identified by a burst of phase encoded signals in a plurality of tracks at the beginning of tape. Such identification was called a TEID. Such tapes were used in conjunction with NRZI tapes, which still required operator intervention for successful recorder operation at multiple lineal densities. The increase of areal densities can also be achieved by increasing track density while not changing lineal density, or by simultaneously changing both densities.

Higher density signal recording often yields an effective higher data rate. This is particularly true with increased lineal densities. When increasing areal densities without increasing the lineal density, the burst data rate does not increase nor does the effective (average) data rate increase. Accordingly, it is desirable to increase track density while simultaneously increasing the effective data rate, all at a minimal cost in the recorder.

As mentioned above, multiple densities in one-half inch tapes required manual selections. With present day automation, it is highly desirable to increase the flexibility of recorders for enabling automatic recorder operation than is amenable to diverse track densities and diverse lineal densities. Many recording devices transfer data signals only a portion of the time; such as rotating head recorders, reciprocating head recorders, during track switching in any recorder, and those recorders employing so-called sector servos. The time remaining in recorder cycles is termed "dead time"--no data is transferred. Such dead time is dedicated for servoing, moving the magnetic medium with respect to a transducer, or vice versa, as well as performing other miscellaneous control functions. It is desirable to minimize the adverse effect on data rates of such dead time in signal recorders. When employing such automatic means, the arrangement on the record media should facilitate the automation of multiple density recording for achieving fully automatic forward/backward compatibility among diverse record members.

It is an object of the invention to enhance signal recorder operation for accommodating diverse densities while enhancing effective signal transfer rates.

It is another object to provide buffering with immediate processing for enhancing data accessability. In accordance with the invention, a preferred record member adapted to be translated in a recording device has a plurality of data signal storage areas. One of the data signal storage areas is always recorded at a lowest possible density--i.e., track or lineal density. Such one data signal storage area contains signals identifying the characteristics of the record member, including track and lineal densities. Others of the data signal storage areas contain signals at track and lineal densities indicated by the one record storage area. The one data signal storage area can be a record medium table of contents area or label area.

A method of operating with signal record members, particularly the record member identified above, includes relatively moving a record member and a relatively wide recording gap transducer a distance less than the width of the transducer for recording narrower tracks. On readback a plurality of narrow readback gaps disposed in an immediate juxtaposition to the relatively wide write gap simultaneously sense a plurality of the recorded narrower tracks for simultaneously supplying plural streams of signals being sensed. One of the streams being sensed is immediately processed, while the other streams of sensed recorded signals are buffered for later processing, such as during dead times of the recorder apparatus. In a rotating head recorder, such dead time is a portion of time when the sensing transducers are not scanning the record member. In the reciprocating head recorder, such dead time can be the flyback time or track switch time between adjacent parallel tracks. In a recording disk, such dead time can be servo sector or track switch times.

A signal readback circuit in accordance with the invention operates with a record member for receiving signals simultaneously from a plurality of read gaps. Signals received via one of the read gaps are immediately supplied to a data utilization means, such as an error correction circuit, while signals from all of the remaining read gaps are buffered awaiting the dead time of the signal recorder. Signal transfer means then transfers the buffered signals from the buffer to the data utilization means during the recorder dead seek or switching time. A programmable control unit is utilized in evaluating signals received from a record medium which indicate track or lineal densities for controlling and selecting recorder operation for accommodating a diversity of track and lineal densities while maintaining a maximal effective signal transfer rate.

Partial buffering of data signals enhances data access while keeping buffer size small.

The foregoing and other objects, features, and advantages of the invention will be apparent from the following more particular description of preferred embodiments of the invention, as illustrated in the accompanying drawings.

FIG. 1 is a diagrammatic showing in the time domain of a magnetic rotating head recorder operation, which advantageously employs the present invention.

FIG. 2 is a simplified diagrammatic showing of a rotating head recorder incorporating the teachings of the present invention.

FIG. 3 is a diagrammatic showing of a record member incorporating the teachings of the present invention and adapted in particular to operate with the FIG. 2 illustrated apparatus.

FIG. 4 is a diagrammatic partial showing of a record storage disk employing the teachings of the present invention.

FIG. 5 shows a magnetic card implementation of the invention.

FIG. 6 diagrammatically shows a disk file with a partial buffer arrangement.

FIGS. 7 and 8 are timing diagrams showing operation of FIG. 6 apparatus.

FIG. 9 is a diagram of a staging storage system having enhanced data access .

Referring now more particularly to the drawings, like numerals indicate like parts and structural features in the various diagrams. A diagrammatic timing diagram with record areas of a record medium usable with a rotating head signal recorder (U.S. Pat. No. 3,932,894) is shown in FIG. 1, Elapsed time from T0 through T1 is a period of the recorder cycle during which a rotating head 16 is scanning a record member (not shown). The portions TP indicate track positioning (servo) signals being sensed. In the illustrated rotating head signal recorder, a track positioning set of signals is located along both edges of the record tape. In proceeding from T0 to T1, the head after scanning TP scans the track identification signals ID, then scans the data area or stripe 10. The particular rotating head signal recorder has a dead time--i.e., a time when the transducer is not scanning the record member from T1 to T0 time. In a pure data transfer sense the dead time occurs not only from T1 to T0 time, but also when the head is scanning TP and ID. Such dead times can be used for processing data signals when such processing is independent of the signals being transferred through a sensing or recording transducer. To this end, doubling the track density in data areas 10 can effectively double the average data rate of the signal recorder without altering the lineal density or the instantaneous burst data rate. Doubling track density is indicated in FIG. 1 by the dashed lines 11 indicating one-half width stripes.

When increasing the track density in this manner, it is desirable to provide forward/backward compatibility between the various data record media. Such forward/backward compatibility and increasing the effective data rate is achieved by the recorder constructed as shown in FIG. 2. It is appreciated that the showing in FIG. 2 is directed toward illustrating the invention and that substantially more circuits and functions are performed in a practical embodiment, all of which functions and circuits are not pertinent to the understanding of the present invention.

FIG. 2 illustrated recorder is a modification of the recorder shown in U.S. Pat. No. 3,932,894. A rotor, indicated by dashed line 15, carries a three gap magnetic transducer 16 for scanning a tape-type record member. Transducer 16 has a write gap 17 with a width W2 equal to the width of original data stripes 10 of FIG. 1. Therefore, during each recording operation, the stripe recorded by the FIG. 2 illustrated recorder will be equal to the original track density of the data recorder. A pair of read gaps 18, 19, having a width W1, are in a trailing relationship to write gap 17. Gaps 18, 19 are adapted to scan essentially one-half the width of an original stripe width 10. In FIG. 1 gap 18 will scan a one-half width stripe 20 simultaneous to gap 19 scanning one-half width stripe 21.

The use of a wide gap to record narrow record tracks (stripes) by a plurality of positioning steps transverse to track length less than the width of a write gap is known. The same technique is used by the present invention in generating half width stripes 20, 21. To this end, data signals received on an input line 25 are supplied to recording circuits 26, which may be constructed in any known manner. Recording circuits 26 supply signals to be recorded to wide write gap 17. The drive 27 which relatively moves the record tape with respect to rotating head 15 is operated in two modes as opposed to the tape positioning mode shown in U.S. Pat. No. 3,932,894. To record at the lower stripe density, whole step flip flop 28 enables drive 27 for recording one stripe to relatively move the tape record medium and the rotating head 16 the width of stripe 10. On the other hand, to record in the higher stripe density to yield stripes 20, 21, for example, half step latch 29 is set to the action condition for conditioning drive 27 for recording one stripe to relatively move the tape record and the rotating head 16 one-half the width of stripe 10. For example, if the movement of the magnetic tape is controlled by a stepping motor, and four steps move the tape with respect to head 16 to achieve a track width 10; then drive 27 provides only two steps to generate one-half width stripes 20, 21 at twice stripe density.

The readback of the lower stripe density exemplified by wide track or stripe 10 is through a single read gap 18. The signals from read gap 19 are ignored. The gap 18 sensed signals are supplied over line 30 to read back circuit 31, such as used in the recorder shown in U.S. Pat. No. 3,932,894. The readback circuit 31 detects the data in the readback signals and supplies a stream of digital data signals over line 32 each time gap 18 scans a data stripe 10. From line 32 the detected digital data signals are supplied to data circuits 33 for further processing. In one constructed embodiment, the data circuit 33 consists of error detection and correction circuits for ensuring that the signals sensed from the record medium are correct and, if not correct, are corrected. Data circuit 33 in turn supplies the detected and corrected signals over line 34 as output signals from the recording system, for example, to CPU or computer 35. In one constructed embodiment, the digital data drive 27 and readback circuit 31 are in a so-called data recording device DRD. The data circuits--i.e., the error correction circuits, and other controls (not shown) associated with a data recording subsystem, are in a storage control unit SCU.

To convert the just described apparatus to double track density, a second readback circuit 37 receives sensed signals from read gap 19. Circuits 31 and 37 operate simultaneously for detecting signals from the gaps 18, 19. During double track density readback, readback circuit 31 continues to supply its detected signals over line 32 to data circuit 33, such as the signals from half width track 20. Simultaneously signals from half width track 21 are detected by readback circuit 37 and supplied as digital data signals over line 38 to a digital data buffer 39. Buffer 39 consists of a random access memory employing semiconductive bistable circuits, magnetic bubbles, CCD's and the like. When gaps 18, 19 reach the trailing end 42 of half width tracks 20, 21, the signals from half width track 20 have been processed by data circuit 33 and the signals from half width track 21 have been stored in buffer 39. At that time a single pole double throw switch 43, of the electronic type, switches data circuit 33 from receiving signals over line 32 to the output terminal 44 of buffer 39. Buffered signals are then automatically transferred (using known techniques) from buffer 39 to be processed by data circuit 33 in the same manner that the signals received from readback circuit 31 were processed. The buffer 39 stored signals are therefore transferred to the output line 34 between times T1 and T0' time while the signals from readback circuit 31 were processed to line 34 between times T0 and T1. In this manner the original burst rate of data signals yields a doubling of the effective data rate of DRD while simultaneously doubling the stripe density.

The various control circuits for effecting the forward/backward compatibilities are described with respect to FIG. 2. Certain signals supplied by data circuit 33 are supplied to CPU 35. Those particular data signals, as will be described later with respect to FIG. 3, indicate whether or not the record medium has double stripe density--i.e., half width stripes 20, 21 or single stripe density corresponding to stripes 10. CPU has programs, not shown nor described, which evaluate the signals for determining the stripe densities. Upon detection of which stripe density is employed, mode select flip flop 46 is set to active or inactive position. If flip flop 46 is in the active condition, the high density H is indicated by electronic switch 47 being set such that common terminal 48 is connected to the H terminal for indicating high density. If, on the other hand, select flip flop 46 is in the inactive condition, switch 47 connects common terminal 48 to the L terminal for indicating low track density.

Switch 47 determines the size of the tape step movements in conjunction with the read/write electronic switch 49. When switch 47 is set to H and switch 49 is set to R, the timed stepping pulses from the step timer 50, which can be the timer shown in U.S. Pat. No. 3,932,894, supplies stepping pulses to whole step circuit 28. On the other hand, if switch 47 is set to L, the timing pulses from step timer 50 actuate whole step circuit 28 for each step. In the high density mode, switch 49 selects either whole step or half step, depending upon the read or write function. During readback, whole steps are used--i.e., since both gaps 18 and 19 simultaneously read the half track width stripes 20, 21, each step can constitute the width of stripe 10. On the other hand, during recording half step circuit 29 actuates drive 27 as previously described to record narrow stripes. The W terminal--i.e., write terminal of switch 49, is also connected to record circuit 26 for activating same. The read/write switch 49, an electronic switch, is activated in accordance with known techniques. For example, when the DRD is connected to a computer, such as CPU 35, a channel control word CCW carries signals through the SCU for selecting the read or write mode. Since such techniques are so well known, they are not further described.

Step timer 50 also controls switch 43 for selectively transferring signals from readback circuit 31 to data circuit 33 or from buffer 39 to data circuit 33. In this regard, step timer 50 supplies its stepping pulse to switch terminal 48 at or near the trailing edge 42 of the data areas 10, 20, 21. In a predetermined time relationship to this stepping pulse, step timer 50 supplies an electronic activating signal, indicated by dashed line 51T, to electronic switch 43 for switching it to terminal 44. Again, when time T0 occurs, step timer 50 supplies a second pulse, using known techniques, returning electronic switch 43 to line 32. Additionally, when switch 47 is set to L, the activating signal from mode select flip flop 46 supplies a gating signal over line 51 to block all switch 43 activations indicated by line 51T. In this manner readback circuit 31 in the low density mode always supplies signals to data circuit 33. Any signals that may be supplid to buffer 39 are ignored.

Alternatively, signals stored in buffer 39 can be used as a backup for signals received from gap 18. That is, if data circuit 33 detects an uncorrectable error in the signals received from half width stripe 20, then the signals stored in buffer 39 from stripe 21, using error recovery procedures beyond the scope of the present invention, replace the stripe 20 supplied signals.

FIG. 3 shows a portion of a tape 55 which employs the present invention. The general format is the same as that shown in U.S. Pat. No. 3,932,894, which includes a so called identification or ID stripe 56, a pair of stripe-indicating tracks or servo tracks 57, a plurality of data stripes 10 and a second pair of stripe indicating or servo tracks 58. The data signal format is related to data signal storage on magnetic recording disks. Many disk storage units have a stack of co-rotating disks; each disk with a multiplicity of concentric record tracks. Axially aligned tracks are termed "cylinders." For a 17 surface unit, each cylinder has 17 axially aligned tracks, one on each surface. The data signal format on the illustrated tape is arranged in blocks of stripes capable of storing one "cylinder" of data. This capacity, with spares, etc., is provided in 67 stripes. Each recording operation must use 67 stripes, one stripe being a guard band. In doubling track density the 67 one-half width stripes occupy 33.5 of the full-width stripes. After recording one cylinder of data, in writing a next adjacent cylinder, the stripe 34' (a full-width stripe) will be partially overwritten, as indicated by dashed line 65. When reading the stripe 34', the A portion will be in the first cylinder whereas the B portion 66 is in the second or next adjacent cylinder. Therefore, the first half width stripe of even numbered cylinders will always begin with a B stripe or the second half of a wide stripe 10.

Returning to FIG. 1, the A half-width stripes correspond to half-width stripe 20 and the B stripes correspond to half-width stripe 21. Of course, if an even number of half stripes were employed for storing a cylinder of data, then each cylinder would start with an A stripe. The "A" portion of stripe 34', as denominated by numeral 67, can contain only special characters--i.e., it is a guard stripe between the cylinder 1 represented by stripes 1A through 33B and the second cylinder beginning with the B stripe 66.

Internal data formats of course can vary and still practice the present invention. For example, areas 62, 63 can contain special data patterns indicating the type of track density employed in that stripe. Therefore, in addition to the volume indicator at 61 each of the data stripes can have a density indicator. For example, two zeros in a row in each of the areas 62, 63 can denominate the low density recording, whereas other bit configurations can indicate an A stripe and a B stripe in the double density configuration. In similar manner, area 56A can be modified to indicate densities, although it is preferred to have the indicators in the data area itself.

From all of the above, it can be seen that the different density recordings can be employed in the same automatic apparatus. That is, a library may contain a plurality of the data cartridges with each data cartridge indicating its own individual densities. This allows migration from a set of low density data cartridges or tape media to a set of higher density data or tape cartridges, wherein the migration occurs during normal operations. This latter feature is another advantage of forward/backward compatibility, particularly in large mass storage systems where data base storage requires constant usage. Of course, the usual changes in support programming in microcode control have to be employed in order to take advantage of the full extent of the forward/backward capabilities.

While the invention has been shown in its best mode as being implemented by a rotating head recorder, no limitation thereto is intended. The invention can be applied with equal facility to diverse recorders. For example, disk and magnetic card recorders can also use the invention to a great advantage. As shown in FIG. 4 a portion of a storage disk 70 includes an outer track 71, which contains CTOC (cartridge table of contents) and an inner track 72 which contains an image of CTOC. Dashed line 73 indicates boundary line between alternating data and servo or track position sectors 74, 75. Accordingly, when a transducer 76 scans the surface of disk 70 for recording and readback via circuit 77, data signals can be transferred during sectors 74 and a dead space occurs during sector 75. Accordingly, the circuits of FIG. 2 can be employed as circuit 77. In this regard, the outer and inner tracks 71, 72 are recorded in the wider track format--i.e., low track density. The intermediate tracks which can be data tracks such as track 78 can be recorded in either of the wide or narrow formats. For example, in the outer area of the disk, the higher speed may enable narrow tracks to be employed. Therefore, track 78 may include two half width tracks 79 and 80 which are operated in the same manner as tracks 20, 21 of FIG. 1. The inner tracks then can contain only the wide tracks. In a similar manner, the lineal densities of track 78 may be increased even though the track width is halved. All of this information is contained in the tracks 71, 72, for track density indicated by the areas 81 in lineal density by areas 82. These areas are repeated around the circumference of the disk for decreasing total initial access time. The disk record of FIG. 4 can also be a spiral track and still employ the teachings of the present invention.

In a similar manner, a magnetic card 100 can be arranged as shown in FIG. 5. Of course instead of concentric circular tracks, there would be parallel longitudinal tracks with a CTOC tracks 101, 102 bracketing the data tracks 104. In this manner the orientation of the card is not as important when inserted into the reader. For example, if card 100 is inserted as from end 92 track 100 would be read first. However, if the card were inserted from end 93, then track 102 would be read first. The recording apparatus can read in either direction provided the detection is symmetrical and the appropriate data handling circuits are provided. Areas 103 contain track and lineal density indicating bits.

Certain aspects of the present invention can be employed independent of multiple track widths; that is, rather than full buffering a storage device, such as a disk file 120 (FIG. 6), the principles of the present invention for real time processing a predetermined portion of signals to or from a disk file while buffering the remaining portion enhances data access while minimizing buffer size, hence cost. In one embodiment, the top one-half disk of disk file 120 a logical portion to be processed in a real time mode while the bottom half portion 121 has signals processed in a buffered mode. The disk surfaces are scanned by the usual transducers with a servo surface guiding the transducer via the usual track seeking and positioning servo mechanism (not shown). In the usual disk file, one and only one transducer is used at a time thereby providing a serial stream of data bits. In the FIG. 6 illustrated embodiment, two transducers are used at a single time to provide two serial strings of data bits respectively transferred over lines 122 and 123 from a transducer (not shown) in the respective halves 119 and 121 of disk file 120. A pair of read-write circuits 124, 125 of usual design are connected respectively to lines 122 and 123. In reading signals from disk file 120, read-write circuits 124 pass its received signals on a real-time basis through switch 130 to cable 131 which is connected to a host or control unit (not shown) in the usual manner. A control line 129 carries signals to actuate switch 130 to pass the signals from read write circuits 124 during a first portion of the read cycle. Switch 130 can be compared to switch 47 of FIG. 2 while line 129 can be compared with terminal 48. Simultaneously, with the real time data transfer from read/write circuits 124, read/write circuits 125 are passing signals from line 123 to buffer 128. Upon completion of the real time data transfer, line 129 signal from a source (not shown) such as a digital computer, actuates switch 130 to connect buffer 128 to cable 131. Then the signals are automatically transferred from buffer 128 over cable 131 using known techniques. While the buffer 128 is transferring the signals, the next track of disk file 120 can be accessed; that is, the seek time is overlapped with the buffer transfer time as best seen in FIG. 7. Arrow 140 represents the real-time data transfer from read/write circuits 124 to cable 131. During the same time period, signals from read/write 125 are transferred to buffer 128 as indicated by arrow 141. The buffer 128 to cable 131 transfer is represented by arrow 142 which is time concurrent with the seek of disk file 120 represented by arrow 143. While the track to track seek time may be substantially longer than the actual buffer signal transfer time, still the reduction in time is significant in a large data processing environment.

Principles of the invention are also applicable to recording signals on disk file 120. In this situation, signals are transferred from cable 131 to disk file 120. Initially, the first signals received are transferred to buffer 128 via switch 130 and as indicated by arrow 145 in FIG. 8. Simultaneously, the head is seeking the appropriate track as represented by arrow 144. Upon completion of the buffer 128 being filled and the seek 144 being completed, signals are recorded in both portions 119 and 121 of disk file 120 as represented by arrows 146 representing the transfer from buffer 128 to read/write circuits 125 and arrow 147 representing the transfer of signals from cable 131 to read/write 124. A control signal on line 129 actuates switch 130 in a known manner for timing the data transfers represented in FIG. 6 through 8.

Another application of the invention is in a staging storage subsystem of FIG. 9. In this instance, a stager such as a staging adapter, IBM Model 3803-3, manufactured by International Business Machines Corporation, Armonk, N.Y., is controlled by a mass storage controller via cable 151 as controlled by the mass storage controller (MSC) (not shown) of the IBM 3850 Mass Storage System, also manufactured by International Business Machines Corporation. Data paths to hosts of stager 150 are represented by double-headed arrow 152. Data access is provided over serial data paths 153 to a staging memory 154 consisting of a plurality of disk files. Signals are generally stored in a data base memory 156 which may be the IBM 3851 Mass Storage Facility (MSF) (not shown) which is a portion of the IBM 3850 Mass Storage System. In general, signals are transferred between staging memory 154 and data base memory 156 over cables 153 and 155 via a buffer (not shown) in stager 150. By using the principles of the present invention, single line cable 153 is replaced by single line cables 122, 123 and the staging memory 154 operates as described with respect to FIG. 6. The practical effect of this arrangement is to free stager 150 for more functions within the mass storage system thereby not only increasing data access of the disk file constituting staging memory 154 but also enables the program stager 150 to provide other mass storage system function in an overlap manner with staging and destaging operations; i.e., transfer of signals between memory 156 and memory 154. For example, if two disk files share stager 150, then the access to the two disk files by stager 150 would be greatly enhanced thereby disproportionately increasing performance of the mass storage system in accordance with the changes effected by the present invention.

Although the present invention has been particularly shown and described with reference to preferred embodiments thereof, it will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention.