Eisi

[0001] The present invention is in the field of error control coding and, in particular, but not exclusively, it relates to the application of error control coding to data for the purposes of ensuring reliable data recovery after the data has been processed in a channel and, for example, stored in a data storage medium or transmitted across a transmission medium.

Background

[0002] As used herein, the term error control coding (BCC) is intended to encompass broadly any kind of coding that is employed to assist in any one or more of: identifying the presence of a data error, identifying where in an amount of data an error is, and correcting an. error. 'Error control coding" should not be confused with the more limited term 'error correction coding', which is treated herein as a sub-set of error control coding.

[0003] in the fields of data storage and data ^' transmission In particular, it i well, known to apply BCC to protect source data before the data is transmitted over a channel to be stored to a data storage medium of transmitted over a transmission medium. The term ' ^' ch&ntteP is intended herein to encompass broadly the means (for example, electronics, optics and any other kinds of data processing elements) by which, for example, data is written to or read from a physical storage medium, or transmitted over or received from a physical or wireless communications medium.

[0004] ECC typically takes the form of additional data (which is sometimes referred to as 'redundant ^* data, or simply 'redundancy'), which is added to or combined with the source data upstream of the channel (that is, before the data is written or transmitted). The term 'source data' is intended herein to mean the data to which the ECC will be applied, and does not imply any particular point of origin, the term 'protected data ^* is intended to mean the source data after it has been protected by ECC, and the term 'recovered data' is intended to mean data that has been retrieved from a storage medium after storage or receipt by a [0005] receiver after transmission. In many cases, but not always, source data and protected data are the same, except that the protected data is accompanied by ECC data when it is applied to a channel. In a perfect system in which no storage or transmission errors occur, it is expected thai the reco vered data should match the source data without the need to employ the ECC data. However, when in a practical, system the ^' recovered data does not match the source data (feat is, data errors have occurred), different degrees of ECC can be used to detect the presence of errors, indicate where the errors are and correct the errors in the recovered data.

[0006] There are a large number of very well known ECC data protection techniques, including parity, checksums, hashing algor thms and Reed-Solomon codes, to name just a few. Commonly, in ECC parlance, a 'distance' is defined as a number of differences (for example, errors) between the .source data and the recovered data. Further, the term 'minimum distance' may be used to indicate, for , given ECC technique, a degree of error detecting and/or correcting that can be achieved; also bearing in mind that some techniques such as Reed-Solomon •coding permi t, for a given minimum distance, a degree of the correctabiiity to be exchanged with a degree of the detectability (and vice versa). The minimum distance can also be thought of as the ^' power, level or degree of protection afforded by the ECC employed,

[0007] Typically, the greater the minimum distance is required to be, the higher the degree of redundancy will be. In other words, for a given number of symbols (for example, bits, bytes, characters or words), which need to be protected using ECC, an. increased minimum distance implies more ECC and an attendant, higher redundancy. Accordingly, system designers are always mindful of the trade-off between how much redundancy ean be tolerated (that is, how efficient a storage system is in terms of storage capacity and/or throughput or how efficient a transmission system is in terms of source data bandwidth) and how well protected the data is required to be (that is, how reliable the system is at recovering from storage or transmission errors). Summary

[0008] According to a first aspect, the present invention provides a data writer, comprising: a memory to store at least one amount of source data thai is to be written to a data storage medium; a processor to arrange the source data into subsets and generate ECC data in respect of each subset, wherein the source data and the associated ECC data are to be written to a. data storage medium via a plurality of individual data channels, and wherein the BCC data, comprises at. least a first degree of ECC protection Iiavirsg a first level of redundancy in respect of a first subset and a second degree of ECC protection having a second level of redundancy in respect of a second subset; a plurality of data writing elements, each, to write data from an associated data channel, concurrently with the writing by the other data writing elements of data from respective data channels, to a data storage medium; and a controller, to control the writing b the data writing elemen ts of the source data and the associated ECC data to the data storage medium.

[0009] Other aspects and embodiments of the present invention will become apparent from the following description and claims,

Brkf De cription,of fe.e_Drawin¾s

[0010] Various features and advantages of the invention will become apparent from the following description of embodiments of the invention, given by way of example only, which is made. with, reference to the accompanying drawings, of which:

FIG. 1. is a schematic diagram representing a short length of magnetic tape medium on which tracks of data have been written;

^' FIG. 2 is a schematic- diagram illustrating an exemplary write head comprising a bank of sixteen write elem nts- of varying width;

PIGs. 3a and 3b are schematic diagrams each representing a two-dimensional array of data, to which ECC has been added in different ways; FIGs. 4a and 4b are schematic diagrams each representing a two-dimensional array of data, to which ECC has been added in different ways;

FIG. 5 is a schematic block diagram representing .major functional components of a tape drive apparatus;

FIG, 6 is a schematic diagram illustrating functionally how an ECC processor generates ECC for plural channels of source data;

FIG. 7 is a. schematic diagram representing an ECC generator associated with one c nnel of an ECC processor;

FIG. 8a is a flow diagram representing the steps involved in a data writing operation; and

FIG. 8b is a flow diagram representing the steps involved . ^' in. a data reading operation.

Detailed Description

[00 Π] Various embodiments of the .present invention will now be described in more detail with reference ^' to the accompanying drawings. It will be appreciated that the invention is not limited in its application to the details of method and the arrangement of components as set forth in the following description or illustrated in the drawings. It will be apparent to a person skilled in the art that additional embodiments of the present invention not detailed in the description are possible and will fall within the scope of the present claims. Accordingly, the. following description should not be interpreted as- limiting in any way, and the scope of protection is defined solely by the claims appended hereto.

[0012] Some embodiments of the present invention relate to data storage, whereby source data to be stored has ECC protection applied to it before the data and ECC are stored to a storage medium such a tape storage medium. According to an embodiment, which will now be described, the tape storage medium is a magnetic tape storage medium,, to which data is written by a magnetic tape storage apparatus. However, it will be appreciated thai other kinds of tape media can be employed, for example optica! tape media to which data is written, by an optica!, tape media apparatus. Indeed, other kinds of storage media can equally be employed, such as (without limitation) magnetic hard disk, magnetic floppy disc or any kind of optical disk (for example, CD, DVD, Bluray, etc.).

[0013] The diagram in FIG. 1 illustrates by way of example only a short section of an elongate magnetic tape storage medium 100, on which, are represented a plurality of elongate tracks of data 105. As depicted, each data track is represented as the space between two parallel black lines. The tracks of data 105 are written along the length of the tape 100 by a write channel comprising a plurality of writing elements 1 10. which are formed in a writing head 11.5 (which is illustrated for exemplary purposes only) of a data writer, which, in this instance, comprises -a tape storage apparatus. This kind of tape storage, in which, data is written in tracks 105 along the length of an elongate tape medium 100, is known as 'linear tape' storage, an example of which is the Uitrium™ tape storage format standard. As shown, the write head 115 has -an array or bank of eight writing elements 1 10, which are shown aligned in the middle of each track. As the tape 100 is wound (typically on reels within, a tape cartridge casing, neithe of which are shown) relative to the head 1.15 by a tape, winding mechanism (also not shown), a group of eight tracks of data can be written substantially concurrently, i parallel. In this case, the channel, in fact, comprises eight channels, each of which handles the writing of a portion of the overall data that is to be stored. Of course, other numbers of tracks can he written instead of eight: for example, four- tracks, sixteen tracks or 32 tracks. The umber of tracks is often (but not necessarily) a power of two, to make data processing in digital data processing environment more straightforward. For the present purposes, each group of concurrently-written tracks will be referred to as a 'track group' 120, which is written to tape in a so-called 'wrap'. Each wrap is referred to as a forward wrap or a reverse wrap, depending on the direction of tape motion relative to the head when the wrap was written. As illustrated in FIG. 1, there are five track groups 120, which are written in five wraps: the first, innermost track group is written in a. first wrap, which is written .from left to right and is referred to as a forward wrap; the next track group is written directly below the inner-most wrap, by a reverse wrap, which is written from right to left; the next track group is written directly above the inner-most track group, and is a forward wrap; the penultimate track group is the lowest group, and is written by a reverse wrap; and the final track group is the uppermost group, which is written by a forward wrap. Of course, there may be a different number of track groups and wraps, with five of each being illustrated by way of example only, it will be appreciated that, at the end of each wrap, the write head is moved in a direction transverse to the longitudinal axis and motion of the tape, in order to position the heads to write each ext track, group.

[0014] When reading the data, which is stored on the tape in FIG. I, a read head with a bank of read elements (not shown), would be employed to read the tracks is the same direction as they were written.

[0015] An important parameter of any data storage system is the storage capacity. Tape formats typically change from one generation to the next: to provide a higher capacity and higher data transfer rates. One way to achieve higher capacity is to employ narrower read and write elements -and space the elements closer together, in order io increase the data density on the storage medium. In addition to problems that arise in manufacturing smaller components, problems can arise in relation to the stability of the storage media. One such problem is tape transverse dimensional stability (TDS). TDS is a measure of changes in tape media width under different environmental conditions. Three factors that affect TDS are environmental conditions the tape media is used in (for example, which depend on temperature and humidity), the tension applied, to the tape media during operation, and aging of the tape media (for example, leading to stretching). As the width of the tracks become narrower, to achieve higher data capacities on tape, controlling TDS, or finding a counter measure io TDS, become critical to successfully designing a product that can operate under an anticipated range of environmental and lifespan expectations, Excessive TDS can lead to track mis-registration, which is a failure to position all parallel reader elements over the written tracks. Embodiments of the present invention provide ways to counter TDS, as will be explained-

[0016] The present in veniors have appreciated thai, in order to meet the physical requirements of TDS, it. is beneficial to make the physical tracks within a single wrap have different widths across a single head 'span', which is the distance between the outer two head elements. According to some embodiments, the elements at the outer edges of the head span axe relativel narrower than the elements at or near the middle of the head span. Typically, in prior art. tape formats, all elements and respective data tracks had substantially the same width.

[0017] In general terms, as track -widths decrease and attendant mechanical tolerances increase, there is an increasing likelihood of read am!/or write errors occurring; One way to compensate for this is to increase the degree of ECC protection applied to the source data before the data is written t tape. However, as has already been indicated, any increase in ECC data reduces storage efficiency, which is undesirable.

[0018] Therefore, according to some embodiments, the present inventors have determined that a varying degree of ECC can be applied to different tracks, in order to maintain acceptable storage efficiency and ECC protection, while still reducing the widths of at leas some of the head elements. More particularly, a greater degree of ECC protection is applied to the relatively narrower tracks while a lesser degree of ECC protection is applied to the relatively wider tracks, in some embodiments, the track widths and respective ECC can be balanced so that, in the aggregate, for a given amount of data to be stored, storage efficiency is not impacted compared with pre-existing formats, as will be described hereafter. [0019] A bead 200 and respective write elements 205 according t© one exemplary embodiment of the. present invention is illustrated in FIG. 2. As shown, the head 200 in this example comprises 16 write elements (associated with tracks numbered 0-15), arranged so that ihe innermost elements (7 and 8) are relatively wide whereas the outermost elements (0 and 15) are relatively .narrow, in FIG. 2, each element is also associated with a width designation, A- E, with A being the widest element and E being the narrowest Elements having the same width designation have substantially the same widths and produce substantially the same width of tracks, as illustrated. As can be seen, elements 8 and 9 (width A) are the widest, and elements 0 and 15 (width E) are the narrowest. Further, elements 1 , 2, 1.3 and .14 (width S are the next narrowest, followed by elements 3.4, 11 and 12 (width€), and then followed by 5, , 9 and 10 (width D). The result of writing data using the head is illustrated in FIG. 2 by the relatively different widths of data track 210, with each track being represented as a shaded band across the page.

[0020] It will be emphasised that the absolute and relative track widths that are shown in FiG.2 are not drawn to scale and that the diagram is merely a depiction of one exemplary embodiment of the invention,

[00211 The diagram in FlGJa illustrates an exemplary set of data 300 comprising a two-dimensional array of protected data, comprising source data bytes 305 and ECC bytes 31 . This example does not constitute an embodiment of the invention and is provided for the purposes of explanation only,

[0022] Source data can be thought of broadly as including any data that is received from a. host computer system or network, which needs to- be stored to tape. Source data may also include format-specific data, which, has been added by upstream processing applied to the received data by a tape storage apparatus, as will be described below. The exemplary data array shown, in FIG. 3a comprises protected source data 'bytes ^* , each comprising 8 data bits. The data array is arranged into subsets, with each subset comprising one of the 16 illustrated rows, each one of which may be written by one write element of an individual channel to a track on a tape storage medium, for example of the kind illustrated in FIG. 1. In this way, each subset, or row, of the data array is written substantially concurrently, and in parallel with the other rows, by the bank of write elements in. the head, In the particular example provided, each row comprises 240 bytes, of which the first 230 bytes are data bytes and the remaining 10 bytes are ECC bytes, in total, the array comprises 3840 bytes, of which 3680 are data bytes and 160 are ECC bytes. It will be appreciated that the dimensions of the array and, more generally, the number of data and ECC bytes, are given by way of example only, and it will be further appreciated that a different data array, with more or fewer data bytes and more or fewer ECC bytes may be provided, fo example, depending on the degree of ECC protection that, is required. It is also to be noted thai the boundaries between bytes hi each row are not illustrated in the diagram in FIG. 3a, for each of illustration only. It will also be appreciated that host data is unlikely to be received in convenient 3680 byte blocks: if there are fewer bytes to store then the tape storage apparatus may add dummy data to pad out the array: if there are more bytes, then, according to some embodiments of the invention, the tape storage apparatus will typicall arrange the data across plural similar data arrays, each .of which is processed and stored to tape in turn.

[0023] According to the example in FIG. 3 a, the ECC bytes for each row of the data array are generated by a Reed-Solomon code generator, which generates a (240, 230, U) coding. In this coding representation, the first number, '240', represents tire total number of bytes, the second number, '230', represents the total, number of source data bytes and the third .number, ' Π ',. represents the amount of ECC protection, in this instance 11, in. terms of the number of bytes that "can _. be corrected. There are other known ways of representing ECC, which could be used instead,

[0024] The present inventors have appreciated that while the data and ECC arrangement of FIG. 3a may be optimal for a write head having write elements of equal width, since each row has associated with it the same number of ECC bytes, the arrangement is not optima! for a write head in which the write elements have, unequal widths, for example as illustrated in FIG. 2. For example, if the data array of FIG. 3a were written by a write head of the kind illustrated in FIG, 2, it may be found that tracks written by a wider element would not benefit from the increased ECC proiectioii, and the tape ibrmai efficiency would suffer from carrying excess redundancy. Likewise, without increasing the ECC protection on the narrow elements, an increase in the amount o -uncorrectable data may arise.

[0025] The diagram in FIG. 3b illustrates one exemplary way of arranging data in. a data array according to an embodiment of the present invention, in which the degree of ECC protection applied to each subset, or row, is generally matched to the characteristics of a respective write element that is expected to write the respective date, to the tape medium. I» this example, the main characteristic which determines the degree of the ECC protection is the width of the respective write element,

[0026] With reference to MG. 3b. the variation in Reed-Solomon coding is applied to each row of the data array, as follows:

- Each of the outer two rows, 0 and 15, is the result of a (240, 226, 14)

Reed-Solomon code applied to 226 data bytes and resulting in 14 ECC bytes.

- Each of rows 1, 2, 13, 14 is the result of a (240, 228, 13) Reed- Solomon code applied to 228 data bytes and resulting in 1.2 ECC bytes.

- Each of rows 3, 4, I I, 12, is the result of a (240, 230, 1 1) Reed- Solomon code applied to 230 data bytes and resulting in 10 ECC bytes. - Each of rows 5, 6, 9, 10, is the result of a (240, 232, 9) Reed- Solomon code applied to 232 bytes of data and resulting in 8 ECC bytes.

Each of rows 7 and 8 is the result of a (240, 234. ?} Reed-Solomon code applied to 234 bytes of data, resulting in. 6 ECC bytes.

[0027 ] Overall, it can he seen thai the data array comprises 3840 bytes, of which 3860 bytes are data bytes and 160 bytes are ECC bytes. As shown, each row has a degree of ECC appropriate for the respective track width. For example: in. one extreme, the narrowest tracks (0 and 15) each have four more ECC bytes than the equivalent tracks in FIG. 3a; and, in the other extreme, the widest tracks (7 and 8) have four fewer ECC bytes than the equivalent tracks in FIG. 3a. In between, tracks 4, 5, 11 and 1 2 have the same degree of ECC as the equivalent tracks in FIG. 3a. Moreover, it will be noted .that the overall number of data bytes and ECC bytes in FIG. 3b matches the number of data bytes and ECC bytes in FIG. 3a. This is because, according to the example in FIG. 3b, the lower degree of ECC applied to the relatively wider tracks is equally balanced by ilie higher degree of ECC applied to the relatively narrower tracks. Of course, there is no reason in principle why such an equal balance is required, and any alternative .arrangement of more or fewer ECC bytes per track, and/or a different size of data array, could be employed as the needs dictate.

[0028} FIG. 4a reproduces FIG. 3b and, in addition, designates selected groups of source dat bytes as V-T and selected groups of ECC bytes as -A ^? -'F. FIG. 4b, illustrates an embodiment in which the identified data bytes and ECC bytes have been re-distributed as follows:

Four ECC bytes hi Group A. row 0 are exchanged with four source data bytes in Group a. row 7

- Two ECC bytes in Group B, row I are exchanged with two source data bytes in Group b, row 6

I I Two ECC bytes in Group C, row 2 are exchanged with two source data bytes in Group c, row 5

- Two ECC bytes in Group D, row 13 are exchanged, with two source data bytes in Group d, row 10

Two ECC bytes in Group B, row 14 are exchanged: with two source data bytes in Group e, row 9

^■■ Four ECC bytes in Group P, row 15 are exchange with two source data bytes in Group f, row 8

{ ^' 0029} As will be evident, after the aforementioned redistribution of source data bytes and ECC bytes, the overall data array resembles the data array that is illustrated in FIG. 3a, insofar as there is an equal number of data bytes and ECC bytes in .each row, A benefit of performing such a redistribution is ^' that the downstream processing, as will be described hereinafter, can be simplified, it will be appreciated that, in reading the data from the tape storage medium to which the data had been written, a respective data reader would be required to reassemble the data and ECC. bytes into their original positions (as shown in FIG. 4a),. before performing airy necessary error detection and correction, to ensure that there is relatively greater degree of ECC applied to data, that was stored in, and retrieved from, the relatively narrower tracks,

[0030] In general, although, in the foregoing examples, the source data bytes are illustrated ^' as being arranged in the left hand side of each ro w and the ECC bytes are illustrated as being arranged in the right hand side of each row, it will be appreciated that such a representation has been provided for ease of explanation only. It is to be emphasised that the source data bytes and ECC bytes may be interleaved, or 'mixed' together, in any manner within each row, or, indeed, among rows. FIG. 4b, for example, illustrates one way in which certain data bytes and ECC bytes are mixed within and between rows. Many other alternative mixes are conceivable. [0031 ] An exemplar)' data writer comprising a linear tape storage apparatus 500, according to an embodiment of the present invention, is illustrated functionally in the diagram in FIG. 5. The apparatus 500 ma he attached directly to a host 502, such as Host A, or indirectly to a host 504, such as Host B, via a local 506 or wide area network 508 such as the Internet. The data writer also acts as a data reader, though, data writing will be discussed in detail in the first instance,

[0032] As shown, the storage apparatus 500 comprises a system control processor 510, which controls the overall operation of the apparatus, either in response to signals received from a user, via a user interface 512, or in response to a software application (not shown), which may be controlling the apparatus 500 .from a host computer 502, 504. The system control processor 510 operates according to program instructions, which are stored in ROM 514 and read therefrom and processed by the control processor 510. The apparatus 500 comprises a data processing pipeline including a host interface 51.6, a format processor 518, an ECC processor 520 and read 522 and write 524 processors... rs a data writing operation, each of these constituent parts process data received from a host to be stored to tape, as will be described. At the end of the pipeline is a write head 526 and a read head 528, which, as shown, each comprise respective (in this example, eight) writing elements 530 and reading elements 532, The motion of the write head 526 is controlled by a head control circuit 534, which in turn is controlled by the system control processor 5.10. Also shown is a tape control circuit 536, which is also controlled by the system control processor 510. The head control circuit 534 and the tape control circuit 536 respectively encompass the mechanical and electrical components that are needed to control the transverse motion of the heads and the forwards and reverse motion of a tape to which data is to he recorded. Finally, the apparatus comprises main memory 538 in the form of ^" RAM for storing data at various stages of processing by the data processing pipeline.

[0033] It will be appreciated that the architecture of a tape storage apparatus according to an embodiment of the present invention may depart from that illustrated in FIG. 5. For example, there amy be fewer or more components employed in any one or more different configurations for perforaiing one or more functions. In some embodiments, the control processor 510 may be an embedded microprocessor or microcontroller and the elements of the pipeline may be embodied in one or more application specific integrated circuits (ASICs), In some embodiments, the Format Processor 518 and the ECC processor 520 may be embodied in a single ASIC, since all data processing is conducted in the digital domain. The write processor 524 meanwhile may be implemented as an ASICs embodying both digital and analogue processing cores, so that digital data can be received and converted into analogue data signals suitable for writing data via write elements to magnetic tape storage. The read processor 522 may also combine digital and analogue capabilities. In any event the particular architecture show is merely illustrative and many other alternative architectures are conceivable,

[0034] The host interface 516 controls the Interactions between the tape storage apparatus 500 and a host, for both data writing and reading operations, in a write operation, the host interface 516 typically receives instructions from the host, the- instructions - being ^' forwarded to the system control processor 510, to control the operation of the apparatus, for example to prepare it for receiving data to be stored. The host interface 516 is also the front end of the pipeline, and manages the transfer of data from the host into the pipeline. The hos Interface typically employs basic, known error detection and detection techniques to ensure that the data which is received is error free. The host interface 516 may employ any one of various known protocols, such as (but not limited to) SCSI, iSCSI, Serial Attached SCSI (SAS), SATA, Fibre Channel, USB, FireWire™.

[0035] The format processor 518 is employed to format the received data, typically according to a proprietary standard, such as the Ultri m tape format standard. For example, the format processor 518 may selectively apply data compression to the received data, using a known data compression format such. as LZW or ALDC. In addition, or alternatively, the format processor 518 may apply encryption to the received data, in order to increase he security of the data once it has been stored to tape. Farther, the format processor 518 may arrange fee data into fixed length blocks or records, for example comprising codewords, the strucmre of which is typically storage-format-specific and independent of the nature of the host data. The particular details of any data storage format are beyond the scope of the present invention, and i is emphasised that a detailed understanding of the nature of any format processor 318, if one is present (and it may not be), is not significant to embodiments of the present invention, if required, further details of the Uiirium 1 data storage format standard are published as ECMA-319 standard "Data Interchange on 12,7 mm, 384-Traek Magnetic Tape Cartridges ··-· UJtriun I Format". Finally, the format processor 518, according to. the present embodiment, arranges the resulting source data Into an array of data bytes, for example as illustrated in FIG. 6, which is stored in the. RAM 538, ready for the. application ofECC by the ECC processor 520, as will now be described.

[0036] An exemplary ECC processor 520 according to the present embodiment is illustrated in FIG. 6. The illustrated ECC processor 520 comprises sixteen individual EGG channel generators: 605, one for each channel, that is each row of source data 610, which is to be written to tape by a respective write element 530. Each ECC channel generator 605 is defined by a generator polynomial function, which operates in a known way on a row of source data 610 that is passed sequentially into the generator, in order to produce the respective ECC bytes 615 that are required to protect the row of data. Obviously, the number of ECC channel-generators may vary. For _. example, instead of providing one ECC channel generator per channel there may be one per amount- of source data so that the same ECC channel generator may be employed consecutively for all rows having the same number of source data bytes. With respect to the data array of FIG. 3b, for example, the same ECC channel generator may be used for rows- i, 2, 13 and 14, as each of those rows has the same number of source data bytes. Likewise, the same ECC channel generator may be used for rows 0 and 15, and a .different ECC channel generator ma be used for rows 3, 4, 11 and 12; and so on. Of course, using the same ECC channel generator for plural rows is likely to slow the storage pipeline down, compared with having an independent ECC channel generator for each channel.

[0037] According to prior art systems in which each row has associated with it an equal number of ECC bytes (or. more generally, ECC symbols), there only needs to be one definition of generator polynomial function to operate on each, row of data. In contrast, embodiments of the present invention typically employ a different generator -polynomial function to produce a respective degree of ECC protection for each respective row of source data. For example, according to an embodiment of the invention, a set of exemplary generator polynomial functions to match the ECC coding requirements of FIG. 3b arc:

(240, 234, 7) : 6 byte ECC

G(x) - <fc+ α (χ±α? ² 6)(χ+ !2 ⁷ )(χ÷

(240, 232, 9) : 8 byte ECC

G.(x) ^~ (x+a ^}24 ){x"i- ^s )(x+ce ^!2S )(x+ ⁷ )(χ+

(.240, 230 J 1} : 10 byte ECC

G(x) = (χ^ ³ ){ + ⁴ )( ^ ^!2 }(χ+ ^)( ^χ ^ ¹²⁷ )(χ+ ^)

(240, 228, 13) ; 12 byte ECC

G(x) = ( ^χ +Φ ² }( ^χ +Φ ^$ )(χ+ a ^m )(x+( ^2S Xx*a }{x+ ¹²⁷ }

(χ+ ^ι ^)(χ^φ ⁹ )(χ+Φ ⁰ )( ^χ ÷ ^ί3! )( ^χ ^Φ ² )( ^χ +Φ ³ )

(240, 226, 14) ; 14 byte ECC (x+w ^2S )(x÷ ^)(x+ a ^!3< >}(x^ where fee r values are the roots of the equations.

[0038] it will be appreciated that the particular generators used would be designed in a known way to suit the particular ECC requirements of the associated channels. In any event, ECC generator polynomial functions may be realised by generator circuits, of which one is exemplified in FIG. 7, by expanding the foregoing equations. For example, the expanded form of the (240, 234, ?) ; 6 byte ECC equation, which is depicted in FIG. 7, is:

G(x) = x ⁶ © a X ^s ® ² x ⁴ ® or ²⁵⁴ * ³ © a ^ZSQ x ² ® a ³⁶ x © 1

[0039 Similar circuits, though with more stages, would he employed to compute the other generator polynomial functions,

[0040] As shown in FIG. 7. a row of source data would be transmitted sequentially into an input 705 of the circuit 700 and would propagate through the circuit. In the example shown, the circuit Is formed as a feedback loop 710, in which there are six register blocks (RG- 5) 715 in series, and an input of each register block is a function of the input values and an. output (where present) o -a previous register block. The inputs to register blocks 1- 5 715 are influenced, in particular, by multiplier blocks 720 each containing an associated expanded coefficient, £z ^ra . Of course, the .number of register blocks 715 and multiplier blocks 72:0 is: adjusted for the degree of ECC protection required. The individually generated ECC bytes are the values that remain in the register blocks RI-R5, after the entire row of source data has been input, and are output (as Outp Q-QmputS) at the end of the process from each respective register block. The manner of generating Reed-Solomon codes using generator polynomial functions as illustrated above and circuits as illustrated in. FIG. 7 is well known in the art of data storage and data transmission, and further detailed explanation will not be provided herein. [0041 ] In addition, in embodiments in which the protected data is re-arranged, for example _* as illustrated in FIG. 4b wherein each row has an equal amount of source data and ECC data, a final step is to re-arrange the data so that each row comprises and equal amount of source data bytes and ECC data bytes.

[0042] it will be appreciated that, although not described herein, in some embodiments of the invention, additional layers or levels of ECC may be employed to protect a set or array of data. For example, in addition to protecting each row of tire array of source data, it ma be desirable to protect each column of the arra of data with additional ECC protection. In such embodiments, for example, the ECC bytes protecting the columns (including columns comprising or including one or more row ECC bytes) may be distributed among the rows before the data is written to tape. Embodiments of the present invention encompass all manner of additional. ECC encoding applied to source data.

[0043] Once the ECC bytes have been generated, and if necessary re-arranged, ^• the protected data (including the source data- bytes and the ECC code bytes) may be written to tape. The protected data may be written directly to tape by the write processor 525 or it may be further encoded by the write -processor, for example by employing known run-length limited (RLE) encoding or similar. Although, not shown, for ease of illustratio only, the write processor typically comprises one processing path for each, channel, to deliver data to each individual write element in the write head. The application and form of any additional coding of the- protected data is beyond the scope of the present invention, in any event, the write processor 524 converts the protected data into a form suitable for energising the write elements to write data to a magnetic tape: medium in a known way.

[0044] A data reading operation, in which the tape apparatus 500 is controlled to read data from a tape medium 100, reverses the flow of data f om the read elements 532, to the read processor 522 (which removes any RLE encoding or the like), through the ECC processor 520 (this time- arranged in a known way to detect and/or correct erroneous data using known ECC recovery techniques). through the format processor 518 (to remove any format specific structure and encoding), and back to a host 502, 504 via the host processor 516. Data read operations; of this kind are generally well known. However, according to embodiments of the present invention, the ECC processor 520 again has to perform error detection and/or correction on rows of data having different amounts of source data. In addition, in embodiments in. which the protected data Is arranged, for example, as illustrated in FIG. 4b, so that each row had an equal amount of source data and ECC data before being stored, a preliminary step is to re-arrange the data so thai the correct ECC data bytes are associated with the respecti e source data rows.

[0045] A data storage operation for storing data according to an embodiment of the present invention will now be described with reference to the flow diagram in FIG. 8a.

[0046] According, to FIG, 8a, in a first step [step 800], a host (for example, 502) instructs a tape storage apparatus.500 according to an embodiment of the present invention to store an amount of data to tape .100. The data is then transferred [step 805] by the host 502, via the host interface 516, into the main memory 538 of the tape storage apparatus 500. Next [step 810]., the format processor 518 optionally applies a format-specific structure to the received data and arranges the data into an array of source- data. An optional step is illustrated hi a flow chart block Bounded by a broken line rather than by a solid line. It will be appreciated that the particular manner in which the data is stored to. main memory 538, at any stage of processing by the tape drive apparatus, is not significant to embodiments of the invention. Next [step 815], the ECC processor 520 generates ECC- data fo each row of the array of source ^' -data and stores: the ECC data, in association with the respective row of source data, n main memory 538. Optionally [step 820], the ECC processor 520 re-arranges the source data and ECC data so that each row in the array of protected data includes the same, number of data bytes and ECC bytes. Next [step 825], the write processor 524 optionally applies RI..L encoding to the data and transmits the data to a respective write element in the write head, for the data to be written to the tape medium [step 830]. As has already been explained, in the event the amount of source data to be stored exceeds the size of one array, plural arrays are generated and stored in series.

[0047] A storage read operation for reading stored data according to an embodiment of the present invention will now be described with reference to the flow diagram in FIG. 8b.

[0048] According to FIG. 8b, in a first step [840), a host (for example, host 502) instructs a tape storage apparatus 500 according to an embodiment of the present invention to read an amount of data from tape 100. The data is read from the tape (step 845] and, optionally [step 850], the read processor 522 removes any RLL encoding. Next [step 855], the ECC processor 510 optionally (if the data and ECC bytes had been re-arranged (for example, balanced) ^' before being ^' stored) re-arranges the recovered data bytes and ECC bytes so that, each row of an associated data array is matched with the appropriate ECC bytes. The ECC processor 520 then [step 860] detects and/or corrects any errors that have arisen in the recovered data. Next [step 865], the for at processor 518 optionally removes any format-specific structure that- has ^' been applied to the data and the ^' host interface 816 returns the data to the requesting host [870].

[0049] It will be appreciated that the examples provided herein of how to arrange data and ECC coding over plural parallel channels represent only a small sample of alternative - arrangements. For example, there is in principle no reason why different degrees of ECC coding cannot be applied to different channels based on channel characteristics other than channel write head width. For example, another channel characteristic may depend on the performance of a ^' write element: if a write element is found to perform slightly less well than other- write heads, the tape apparatus may b adaptable to apply additional ECC to the channel thai is written by the respective write element. In another example, if the tape (or other storage medium) to which the data is written is found to be less reliable in the longitudinal direction along the direction of one particular write element, for example disc to wear or a scratch, then additional BCC protection may be applied to the channel that is to be written over that area of tape. Of course, if the tape is no longer of an acceptable standard, it would not be used at all,

[0050] While embodiments of the present invention have been mainly described by reference to data storage, there is no reason why the same principle, of applying different degrees of BCC protection to different channels, cannot be employed in communications systems in which data is communicated, substantially in parallel, over plural separate channel, for example in dependence upon the nature or characteristics of the respective channels. Embodiments of the present invention, therefore, encompass data transmission and reception as well as data storage and recover,'.

[0051] The above, embodiments are to be understood as illustrative examples of the. invention. Further embodiments of the invention are envisaged. For example, similar principles ma be applied to other kinds of magnetic storage media, such as floppy disk or hard disk, and to other kinds of data writer or data reader, for example using optical rather than magnetic storage techniques. It is to be understood that any feature described in relati n to any one embodiment may be used alone, or, if the context permits, in combination with other features described, and. may also be used in combination with one or -more features of any other of the embodiments, or any combination of any other of the embodiments. Furthermore, equivalents and modifications not described above may also be employed without departing from the scope of the invention, which is defined in the accompanying claims.