WO2003063161A2

WO2003063161A2 - Data storage system, file system and method for handling data on a data storage apparatus

Info

Publication number: WO2003063161A2
Application number: PCT/IB2002/005631
Authority: WO
Inventors: Rudi J. M. Wijnands; Lambert H. A. Jacobs
Original assignee: Koninklijke Philips Electronics N.V.
Priority date: 2002-01-24
Filing date: 2002-12-19
Publication date: 2003-07-31
Also published as: WO2003063161A3

Abstract

In a file system, it might not be possible to write a file to a disc without encountering defects. By having a reserved area on a data storage medium assigned to each allocation unit of the file system, it is possible to slip defects permanently. Physically slipping defects is a known concept inside hard discs and is applied during manufacturing. New defects cannot be slipped inside the hard disc, but the concept is applied according to the invention in the file system. Applying the concept of defect skipping and slipping in a file system allows to skip and slip defects in the file system and assures that new defects are never accessed any more in the future. Such measure is desirable to prevent real-time performance penalties.

Description

DATA STORAGE SYSTEM, FILE SYSTEM AND METHOD FOR HANDLING DATA ON A DATA STORAGE APPARATUS

The invention regards a data storage system comprising a data storage apparatus having a data storage medium with a plurality of extended allocation units, further comprising a file system having a plurality of logical allocation units respectively assigned to the extended allocation units. Further the invention regards a file system storable on a readable medium having a plurality of logical allocation units respectively assigned to extended allocation units on a data storage medium. The invention also regards a method for handling data on a data storage apparatus comprising a data storage medium, having a user area and a spare area, wherein upon detection of a defect sector on the data storage apparatus the data assigned to the defect are assigned to the next non-defective sector and other data are allocated, if necessary, into a reserved sector.

Hard disc based devices recording multi-media streams like mpeg-encoded video require a real-time file system for writing the data to a disc and for reading the data back. Conventional real-time file systems may try to write all files in time, but sometimes do not succeed, for example because of disc problems. There are then two options: writing the data too late or discarding some of the unwritten data. Both options may lead to a significant data loss. In the first option one might loose data from the point where the error occurred to the end of the total request. In the second option one would loose only a part, but one might successfully read/write the last part of the request. Conventional data oriented operating systems have no real-time requirements and therefore usually tend to aim for maximum data integrity delaying completion of each command until properly executed. Therefore, conventional data storage systems show, especially upon occurrence of defects on a data storage medium, substantial loss of performance as they are not adapted for real-time requirements. Defects are as far as possible avoided or fixed. When a defect is detected during reading, the hard disc will try to use very extensive error recovery algorithms to retrieve the original data. If successful in doing so, it will perform a media test. A media test is more or less a number of write- verify actions. If all write- verify actions are successful, it will leave the data in the original sector and treat it as a normal non-defective sector. If one of the write- erify actions fails, it will reallocate the defective sector to a spare area and rewrite the original data at the location. If the hard disc cannot retrieve the original data with the extensive error recovery, then it will be report this to its host and mark the sector to be a pending defect. It is more or less up to the host to decide what to do. Next time when to host requests to write to this pending defect, the hard disc will try to write the data and verify a number of times. If each write-verify action in the hard disc is successful, then the pending defect is treated as a normal non-defective sector. If not, then the pending defect is internally changed to a permanent defect and the hard disc will reallocate the sector to a spare area and write the data there. In a conventional hard disc there will always be enough time, because the hard disc will always use as much time as necessary to try to solve errors. In a hard disc prepared with an audio/video streaming feature set there will be a command completion time limit. This means the hard disc has limited time for error recovery, meaning that it possibly cannot solve the error in time. A possible cause of such an error may be a defect, which is a physically damaged sector on the hard disc medium. Even with extensive error recovery the hard disc will never be able to repair such sector. In current file systems the smallest addressable unit is called an allocation unit.

Files are built from one or more allocation units, depending on the amount of data that needs to be written. hi particular manufacturing defects on hard discs are fixed by skipping a defect sector and slipping the good sectors before the hard disc drive system leaves the fab. In current systems however, this can be done only in the fab. When an allocation unit with the defect sector is read the hard disc system has to swap or jump to the data allocated in a preferably remote spare area for reading such data. This leads to a non-usable allocation unit for a real-time file system. This is particular true when a file-system is not provided with information on sectors allocated in a spare area and does not take a performance penalty into account. In conventional file system this is not noticed by the host of the hard disc. Reallocations of sectors are a hard disc internal matter. The reallocation only causes a performance penalty, which in a conventional Personal Computer (PC) application does not matter at all. However, in a real-time streaming application that uses a real-time file system the real-time file system needs predictable behavior of the hard disc. Since a reallocation is not 'visible' to the host, this means such a real time file system will not know about the performance penalty. As a result, a command to the hard disc will take longer than expected, which in turn can lead to buffer over- or underflows.

In the US 6,101,619 a scheme is provided to reduce the number of searches for such data by accessing replaced sectors at preferably later times subsequent to a usual data access, the US 5,166,936 or the WO 98/03970 low level formatting of tracks is suggested to build good tracks of data and to prevent a further access of a defect. Such measures take considerable effort and may only be done in idle time. Moreover, such transaction schemes should be guarded against power failures. Such schemes require for a track switch as they rely on remote spare areas as in the US 6,201,655 Bl.

In the US 6,034,831 it is suggested to write data, which were originally scheduled to be written to a defective region, to a substantially adjacent non-defective area on the original track, if the size of the defective region does not exceed a threshold size, and to write such data to a remote non-defective area outside the original track if the size of the defective region exceeds a threshold size. This scheme still requires for a track switch if the defect size exceeds a threshold size. Such threshold size is preferably only on the amount of one sector. Therefore, the scheme is restricted to an extremely limited threshold size of defect data. All conventional schemes require a swap or jump of the data transfer head to another track inducing a performance penalty. Therefore, such schemes are not be able to guarantee a request service time in case of an access to a defective region that contains a defect sector. This may result either in a degradation of performance or a loss of data.

This is where the invention comes in, the object of which is to specify a data storage system, comprising a data storage medium and a file system and further a file system and further a method for handling data on a data storage apparatus adapted such that a data request is more effective and in particular to avoid huge data loss and allocation space loss due to defects even in case of an access to a region of the storage medium containing defective or replaced sectors.

The object regarding the data storage system is solved by the data storage system mentioned in the introductory, wherein according to the invention at least for a significant number of allocation units the number of sectors assigned to the extended allocation unit exceeds the number of sectors assigned to the related logical allocation unit. The significant number may extend 10% of the allocation units and may be at least the majority of allocation units, hi a particular preferred configuration for each allocation unit the number of sectors assigned to the extended allocation unit exceeds the number of sectors assigned to the related logical allocation unit. Thereby it is possible to slip defects permanently.

The invention has arisen from the desire to avoid major disc space loss due to defects in allocation units of the file system. Especially in real-time file system the allocation unit size will be typically like 10 MB or larger. In case an allocation unit contains a defect, a file system conventionally would discard this allocation unit from its allocation space, if possible at the time. This means that 10 MB would become unusable, because one single sector of 512 bytes is damaged. The main insight of the proposed invention is, that in case the defect slipping is done in the file system, such a defect sector will not be addressed anymore. As a result a proposed data storage system will in general not suffer from a performance penalty anymore. However, a first time when a defect is detected one still may have a performance penalty, as additional administration may be introduced to the file system to internally keep track of a detected defect and slipped sectors in an allocation unit. Conventionally even if data originally scheduled to a defect region are allocated to a remote or other spare region this may cause significant performance losses as still a jump or swap of a data transfer head is necessary to access such replaced sectors. The proposed data storage system is capable to define an extended allocation unit which is larger than a logical allocation unit and wherein the extended allocation unit comprises the full logical allocation unit. Thereby a file system or a real-time file system may be established which is able to provide efficient defect slipping in the file system. The size of the extended allocation unit is determined by the file system. The relatively small loss of useable data space to be sacrificed is considered to be of minor interest, compared to the loss in case of defects in a conventional file system with large allocation units. In a further developed data storage system advantageously the extended allocation unit comprises a number of user sectors being part the user area and wherein the number of user sectors correspond to the size of the logical allocation unit and which further comprises a number of reserved sectors, in particular being part of an user area.

Further a data apparatus advantageously comprises a data transfer head, a drive to rotate the medium and a servo to move the head, further a controller having a control electronics, a microprocessor and a memory and further an interface for connecting the disc drive to a host.

Also the invention leads to a file system as mentioned in the introductory by which the object regarding the file system is solved and wherein a logical allocation unit assigns a number of user sectors of a user area and a number of reserved sectors of a user area in the respective extended allocation unit.

The object regarding the method is solved by a method for handling data on a data storage apparatus comprising a data storage medium, in particular by a method for handling data on a disc drive comprising a data storage disc, having a user area, wherein upon detection of a defect sector on the data storage apparatus the data assigned to the defect are assigned to the next non-defective sector and other data are allocated, if necessary, into a reserved sector in particular in the user area, wherein according to the invention in a file system the defect and reserved sector are assigned to the same logical allocation unit. In a preferred embodiment of the method the logical allocation unit in the file system is assigned to an extended allocation unit on the data storage medium, wherein the extended allocation unit contains a number of user sectors being part of the user area and a number of reserved sectors, in particular being also part of the user area.

The invention will now be described with reference to the accompanying drawing. The figures of the drawing illustrate in a schematic and not necessarily scaled form preferred embodiments of the invention compared to prior art. The figures illustrate in: Figure la: a scheme of sector skipping and slipping;

Figure lb: allocation and mapping of a defective sector due to a grown defect into a spare area;

Figure 2: a preferred embodiment of the invention providing a reserved area array assigned to essentially each of a plurality of allocation units.

Conventionally only during manufacturing defective sectors are skipped as shown in Figure la. According to the preferred embodiment a defective sector e. g. with physical block address 3 (PBA 3) is replaced by the next immediate sector in order to maintain the sequential ordering of logical data sequences. This technique eliminates the need to seek to another track to access a replacement of an sector allocated in a remote spare area. If defects, known as grown defects, occur during application of a hard disc drive, such skip and slip scheme is possible in the embodiment within a wide and unlimited range in the file system, as reserved sectors are provided for essentially each logical allocation unit in an extended allocation unit of the file system.

Conventionally defects that occur during application are, if found, allocated to a remote spare sector. In the situation depicted in Figure lb, the physical sector PBA 3 is allocated to the replacement sector S2 in a remote spare area.

Defects cause in general a loss of valid data space being of the size of one allocation unit. A file system typically removes a whole allocation unit from its allocation space, if it contains a defect. Therefore, as outlined in Figure 2 a preferred embodiment of the invention provides to increase the size of an allocation unit, but keeping it logically the same size. In particular an extended allocation unit 1 is favorable in which e. g. sectors #1 to #13 are assigned to a user area UA and are accompanied by further reserved sectors #14 to #17 as assigned to a reserved area RA to build one extended allocation unit 1. Therefore, in this exemplifying embodiment one allocation unit comprises thirteen user sectors #1 to #13 and four reserved sectors #14 to #17. The reserved sectors are also located in the user area. If a defect should occur during use of a hard disc drive in most cases such allocation unit 1 is adapted such that the defect sector is skipped and all following sectors are slipped into the reserved area RA of the extended allocation unit 1. Thereby it is possible to introduce the skip and slip scheme in the file system. The specified representation of an allocation unit in Figure 2 does not indicate the number of user sectors and the number of reserved sectors in a user area UA and a reserved area RA of an extended allocation unit according to this embodiment in general. The size of an allocation unit is dependent on the file system. The number of reserved sectors in addition to the number of sectors corresponding for the logical size may be any arbitrary number suitable for such file system and suitable to perform a skip and slip scheme in an effective way. hi particular it has to be accounted for that increasing the number of reserved sectors of a reserved area RA added to the extended allocation unit 1 may result in that more defects per allocation unit may be skipped and slipped. Also the loss of usable data space is increased with the increase of a reserved area RA per extended allocation unit 1. Nevertheless such loss will in general be smaller than the loss in case of defects in a file system with large allocation units. The size of an allocation unit maybe of multiple megabytes like 10 MB or even larger.

The size of an allocation unit may be increased to a suitable size. Further only a limited logical size of such an allocation unit i. e. in general the same logical size as originally intended for data allocation, may be provided. On the other hand, in case a defect occurs, additional sectors of an enlarged reserved area RA serve as a reserve to perform a skip and slip of the defects possibly located in the original data space i. e. in the user area UA of the allocation unit 1. Further administration in the file system may be introduced to internally keep track of these slipped sectors in an allocation unit. Such administration e. g. may be provided by a protocol between the controller of a hard disc drive and a host.

While there has been shown and described what is considered to be preferred embodiments of the invention it will of course be understood that various modifications and changes in form or detail could readily be made without departing from the spirit of the invention. It is therefore intended that the invention may not be limited to the exact form or detail herein shown and described nor to anything less than the whole of the invention herein disclosed and as herein after claimed.

The invention can be summarized as follows: In a file system, it might not be possible to write a file to a disc without encountering defects. By having a reserved area on a data storage medium assigned to each allocation unit of the file system, it is possible to slip defects permanently. Physically slipping defects is a known concept inside hard discs and is applied during manufacturing. New defects cannot be slipped inside the hard disc, but the concept is applied according to the invention in the file system. Applying the concept of defect skipping and slipping in a file system allows to skip and slip defects in the file system and assures that new defects are never accessed any more in the future. Such measure is desirable to prevent real-time performance penalties.

Claims

CLAIMS:

1. Data storage system comprising a data storage apparatus having a data storage medium with a plurality of extended allocation units (1), comprising a file system having a plurality of logical allocation units respectively assigned to the extended allocation units (1), characterized in that at least for a significant number of allocation units the number of sectors (#1 - #17) assigned to the extended allocation unit (1) exceeds the number of sectors (#1 - #13) assigned to the related logical allocation unit.

2. Data storage system as claimed in claim 1, characterized in that the extended allocation unit (1) comprises a number of user sectors (#1 - #13) being part of a user area (UA) and wherein the number of user sectors (#1 - #13) corresponds to the size of the logical allocation unit and the extended allocation unit (1) further comprises a number of reserved sectors (#14 - #17), in particular being also part of a user area (UA).

3. Data storage system as claimed in claim 1 or 2, characterized in that the significant number is the number of all allocation units.

4. File system storable on a readable medium having a plurality of logical allocation units respectively assigned to extended allocation units (1) on a data storage medium characterized in that a logical allocation unit assigns a number of user sectors (#1 - #13) of a user area (UA) and a number of reserved sectors (#14 - #17), of a user area, in particular in the respective extended allocation unit (1).

5. Method for handling data on a data storage apparatus comprising a data storage medium, in particular for handling data on a disc drive comprising a data storage disc, having a user area (UA), wherein upon detection of a defect sector on the data storage apparatus the data assigned to the defect are assigned to the next non-defective sector and other data are allocated, if necessary, into a reserved sector, in particular in the user area (UA), characterized in that in a file system the defect and reserved sector are assigned to the same logical allocation unit.

6. Method as claimed in claim 5, characterized in that the logical allocation unit in the file system is assigned to an extended allocation unit (1) on the data storage medium, wherein the extended allocation unit (1) contains a number of user sectors (#1 - #13) being part of the user area (UA) and a number of reserved sectors (#14 - #17), in particular being also part of the user area.