Level 1 provides much better write performance than level 5
● Level 5 requires at least 2 block reads and 2 block writes to write a
single block, whereas Level 1 only requires 2 block writes
● Level 1 preferred for high update environments such as log disks
■ Level 1 had higher storage cost than level 5
● disk drive capacities increasing rapidly (50%/year) whereas disk
access times have decreased much less (x 3 in 10 years)
● I/O requirements have increased greatly, e.g. for Web servers
● When enough disks have been bought to satisfy required rate of I/
O, they often have spare storage capacity
so there is often no extra monetary cost for Level 1!
90 trang |
Chia sẻ: huyhoang44 | Lượt xem: 827 | Lượt tải: 0
Bạn đang xem trước 20 trang tài liệu Cơ sở dữ liệu - Chapter 11: Storage and file structure, để xem tài liệu hoàn chỉnh bạn click vào nút DOWNLOAD ở trên
Database System Concepts, 5th Ed.
©Silberschatz, Korth and Sudarshan
See www.dbbook.com for conditions on reuse
Chapter 11: Storage and File Structure
Rev. Aug 1, 2008
©Silberschatz, Korth and Sudarshan11.2Database System Concepts 5th Edition
Chapter 11: Storage and File Structure
n Overview of Physical Storage Media
n Magnetic Disks
n RAID
n Tertiary Storage
n Storage Access
n File Organization
n Organization of Records in Files
n DataDictionary Storage
©Silberschatz, Korth and Sudarshan11.3Database System Concepts 5th Edition
Classification of Physical Storage Media
n Speed with which data can be accessed
n Cost per unit of data
n Reliability
l data loss on power failure or system crash
l physical failure of the storage device
n Can differentiate storage into:
l volatile storage: loses contents when power is switched
off
l nonvolatile storage:
Contents persist even when power is switched off.
Includes secondary and tertiary storage, as well as
batterybacked up mainmemory.
©Silberschatz, Korth and Sudarshan11.4Database System Concepts 5th Edition
Physical Storage Media
n Cache – fastest and most costly form of storage; volatile;
managed by the computer system hardware
l (Note: “Cache” is pronounced as “cash”)
n Main memory:
l fast access (10s to 100s of nanoseconds; 1 nanosecond =
10–9 seconds)
l generally too small (or too expensive) to store the entire
database
capacities of up to a few Gigabytes widely used
currently
Capacities have gone up and perbyte costs have
decreased steadily and rapidly (roughly factor of 2
every 2 to 3 years)
l Volatile — contents of main memory are usually lost if a
power failure or system crash occurs.
©Silberschatz, Korth and Sudarshan11.5Database System Concepts 5th Edition
Physical Storage Media (Cont.)
n Flash memory
l Data survives power failure
l Data can be written at a location only once, but location can
be erased and written to again
Can support only a limited number (10K – 1M) of
write/erase cycles.
Erasing of memory has to be done to an entire bank of
memory
l Reads are roughly as fast as main memory
l But writes are slow (few microseconds), erase is slower
©Silberschatz, Korth and Sudarshan11.6Database System Concepts 5th Edition
Physical Storage Media (Cont.)
n Flash memory
l NOR Flash
Fast reads, very slow erase, lower capacity
Used to store program code in many embedded devices
l NAND Flash
Pageatatime read/write, multipage erase
High capacity (several GB)
Widely used as data storage mechanism in portable
devices
©Silberschatz, Korth and Sudarshan11.7Database System Concepts 5th Edition
Physical Storage Media (Cont.)
n Magneticdisk
l Data is stored on spinning disk, and read/written magnetically
l Primary medium for the longterm storage of data; typically stores
entire database.
l Data must be moved from disk to main memory for access, and
written back for storage
l directaccess – possible to read data on disk in any order,
unlike magnetic tape
l Survives power failures and system crashes
disk failure can destroy data: is rare but does happen
©Silberschatz, Korth and Sudarshan11.8Database System Concepts 5th Edition
Physical Storage Media (Cont.)
n Optical storage
l nonvolatile, data is read optically from a spinning disk using
a laser
l CDROM (640 MB) and DVD (4.7 to 17 GB) most popular
forms
l Writeone, readmany (WORM) optical disks used for
archival storage (CDR, DVDR, DVD+R)
l Multiple write versions also available (CDRW, DVDRW,
DVD+RW, and DVDRAM)
l Reads and writes are slower than with magnetic disk
l Jukebox systems, with large numbers of removable disks, a
few drives, and a mechanism for automatic loading/unloading
of disks available for storing large volumes of data
©Silberschatz, Korth and Sudarshan11.9Database System Concepts 5th Edition
Physical Storage Media (Cont.)
n Tape storage
l nonvolatile, used primarily for backup (to recover from disk
failure), and for archival data
l sequentialaccess – much slower than disk
l very high capacity (40 to 300 GB tapes available)
l tape can be removed from drive ⇒ storage costs much
cheaper than disk, but drives are expensive
l Tape jukeboxes available for storing massive amounts of
data
hundreds of terabytes (1 terabyte = 109 bytes) to even a
petabyte (1 petabyte = 1012 bytes)
©Silberschatz, Korth and Sudarshan11.10Database System Concepts 5th Edition
Storage Hierarchy
©Silberschatz, Korth and Sudarshan11.11Database System Concepts 5th Edition
Storage Hierarchy (Cont.)
n primary storage: Fastest media but volatile (cache, main
memory).
n secondary storage: next level in hierarchy, nonvolatile,
moderately fast access time
l also called online storage
l E.g. flash memory, magnetic disks
n tertiary storage: lowest level in hierarchy, nonvolatile, slow
access time
l also called offline storage
l E.g. magnetic tape, optical storage
©Silberschatz, Korth and Sudarshan11.12Database System Concepts 5th Edition
Magnetic Hard Disk Mechanism
NOTE: Diagram is schematic, and simplifies the structure of actual disk drives
©Silberschatz, Korth and Sudarshan11.13Database System Concepts 5th Edition
Magnetic Disks
n Readwrite head
l Positioned very close to the platter surface (almost touching it)
l Reads or writes magnetically encoded information.
n Surface of platter divided into circular tracks
l Over 50K100K tracks per platter on typical hard disks
n Each track is divided into sectors.
l Sector size typically 512 bytes
l Typical sectors per track: 500 (on inner tracks) to 1000 (on outer
tracks)
n To read/write a sector
l disk arm swings to position head on right track
l platter spins continually; data is read/written as sector passes
under head
©Silberschatz, Korth and Sudarshan11.14Database System Concepts 5th Edition
Magnetic Disks (Cont.)
n Headdisk assemblies
l multiple disk platters on a single spindle (1 to 5 usually)
l one head per platter, mounted on a common arm.
n Cylinder i consists of ith track of all the platters
n Earlier generation disks were susceptible to “headcrashes”
leading to loss of all data on disk
l Current generation disks are less susceptible to such
disastrous failures, but individual sectors may get corrupted
©Silberschatz, Korth and Sudarshan11.15Database System Concepts 5th Edition
Disk Controller
n Disk controller – interfaces between the computer system and
the disk drive hardware.
l accepts highlevel commands to read or write a sector
l initiates actions such as moving the disk arm to the right track
and actually reading or writing the data
l Computes and attaches checksums to each sector to verify
that data is read back correctly
If data is corrupted, with very high probability stored
checksum won’t match recomputed checksum
l Ensures successful writing by reading back sector after writing
it
l Performs remapping of bad sectors
©Silberschatz, Korth and Sudarshan11.16Database System Concepts 5th Edition
Disk Subsystem
n Disk interface standards families
l ATA (AT adaptor) range of standards
l SATA (Serial ATA)
l SCSI (Small Computer System Interconnect) range of
standards
l Several variants of each standard (different speeds and
capabilities)
©Silberschatz, Korth and Sudarshan11.17Database System Concepts 5th Edition
Performance Measures of Disks
n Access time – the time it takes from when a read or write request
is issued to when data transfer begins. Consists of:
l Seek time – time it takes to reposition the arm over the correct
track.
Average seek time is 1/2 the worst case seek time.
– Would be 1/3 if all tracks had the same number of
sectors, and we ignore the time to start and stop arm
movement
4 to 10 milliseconds on typical disks
l Rotational latency – time it takes for the sector to be accessed
to appear under the head.
Average latency is 1/2 of the worst case latency.
4 to 11 milliseconds on typical disks (5400 to 15000 r.p.m.)
©Silberschatz, Korth and Sudarshan11.18Database System Concepts 5th Edition
Performance Measures (Cont.)
n Datatransfer rate – the rate at which data can be retrieved from
or stored to the disk.
l 25 to 100 MB per second max rate, lower for inner tracks
l Multiple disks may share a controller, so rate that controller can
handle is also important
E.g. ATA5: 66 MB/sec, SATA: 150 MB/sec, Ultra 320 SCSI:
320 MB/s
Fiber Channel (FC2Gb): 256 MB/s
©Silberschatz, Korth and Sudarshan11.19Database System Concepts 5th Edition
Performance Measures (Cont.)
n Mean time to failure (MTTF) – the average time the disk is
expected to run continuously without any failure.
l Typically 3 to 5 years
l Probability of failure of new disks is quite low, corresponding to
a theoretical MTTF of 500,000 to 1,200,000 hours for a new
disk
E.g., an MTTF of 1,200,000 hours for a new disk means that
given 1000 relatively new disks, on an average one will fail
every 1200 hours
l MTTF decreases as disk ages
©Silberschatz, Korth and Sudarshan11.20Database System Concepts 5th Edition
Optimization of DiskBlock Access
n Block – a contiguous sequence of sectors from a single track
l data is transferred between disk and main memory in
blocks
l Typical block sizes today range from 4 to 16 kilobytes
n Diskarmscheduling algorithms order pending accesses to
tracks so that disk arm movement is minimized
l elevator algorithm : move disk arm in one direction (from
outer to inner tracks or vice versa), processing next request
in that direction, till no more requests in that direction, then
reverse direction and repeat
©Silberschatz, Korth and Sudarshan11.21Database System Concepts 5th Edition
Optimization of Disk Block Access (Cont.)
n File organization – optimize block access time by organizing
the blocks to correspond to how data will be accessed
l E.g. Store related information on the same or nearby
blocks/cylinders.
File systems attempt to allocate contiguous chunks of
blocks (e.g. 8 or 16 blocks) to a file
l Files may get fragmented over time
E.g. if data is inserted to/deleted from the file
Or free blocks on disk are scattered, and newly created
file has its blocks scattered over the disk
Sequential access to a fragmented file results in
increased disk arm movement
l Some systems have utilities to defragment the file system,
in order to speed up file access
©Silberschatz, Korth and Sudarshan11.22Database System Concepts 5th Edition
n Nonvolatile write buffers speed up disk writes by writing blocks to
a nonvolatile RAM buffer immediately
l Nonvolatile RAM: battery backed up RAM or flash memory
Even if power fails, the data is safe and will be written to disk
when power returns
l Controller then writes to disk whenever the disk has no other
requests or request has been pending for some time
l Database operations that require data to be safely stored
before continuing can continue without waiting for data to be
written to disk
l Writes can be reordered to minimize disk arm movement
Optimization of Disk Block Access (Cont.)
©Silberschatz, Korth and Sudarshan11.23Database System Concepts 5th Edition
n Log disk – a disk devoted to writing a sequential log of block
updates
l Used exactly like nonvolatile RAM
Write to log disk is very fast since no seeks are required
No need for special hardware (NVRAM)
n File systems typically reorder writes to disk to improve
performance
l Journaling file systems write data in safe order to NVRAM
or log disk
l Reordering without journaling: risk of corruption of file system
data
Optimization of Disk Block Access (Cont.)
©Silberschatz, Korth and Sudarshan11.24Database System Concepts 5th Edition
RAID
n RAID: Redundant Arrays of Independent Disks
l disk organization techniques that manage a large numbers of
disks, providing a view of a single disk of
high capacity and high speed by using multiple disks in
parallel, and
high reliability by storing data redundantly, so that data can
be recovered even if a disk fails
n The chance that some disk out of a set of N disks will fail is much
higher than the chance that a specific single disk will fail.
l E.g., a system with 100 disks, each with MTTF of 100,000
hours (approx. 11 years), will have a system MTTF of 1000
hours (approx. 41 days)
©Silberschatz, Korth and Sudarshan11.25Database System Concepts 5th Edition
Improvement of Reliability via Redundancy
n Redundancy – store extra information that can be used to
rebuild information lost in a disk failure
n E.g., Mirroring (or shadowing)
l Duplicate every disk. Logical disk consists of two physical
disks.
l Every write is carried out on both disks
Reads can take place from either disk
l If one disk in a pair fails, data still available in the other
Data loss would occur only if a disk fails, and its mirror
disk also fails before the system is repaired
– Probability of combined event is very small
» Except for dependent failure modes such as fire or
building collapse or electrical power surges
©Silberschatz, Korth and Sudarshan11.26Database System Concepts 5th Edition
Improvement of Reliability via Redundancy
n Mean time to data loss depends on mean time to failure,
and mean time to repair
l E.g. MTTF of 100,000 hours, mean time to repair of 10
hours gives mean time to data loss of 500*106 hours (or
57,000 years) for a mirrored pair of disks (ignoring
dependent failure modes)
©Silberschatz, Korth and Sudarshan11.27Database System Concepts 5th Edition
Improvement in Performance via Parallelism
n Two main goals of parallelism in a disk system:
1. Load balance multiple small accesses to increase throughput
2. Parallelize large accesses to reduce response time.
n Improve transfer rate by striping data across multiple disks.
n Bitlevel striping – split the bits of each byte across multiple disks
l But seek/access time worse than for a single disk
Bit level striping is not used much any more
n Blocklevel striping – with n disks, block i of a file goes to disk (i
mod n) + 1
l Requests for different blocks can run in parallel if the blocks
reside on different disks
l A request for a long sequence of blocks can utilize all disks in
parallel
©Silberschatz, Korth and Sudarshan11.28Database System Concepts 5th Edition
RAID Levels
n RAID organizations, or RAID levels, have differing cost,
performance and reliability characteristics
n RAID Level 0: Block striping; nonredundant.
l Used in highperformance applications where data lost is not
critical.
n RAID Level 1: Mirrored disks with block striping
l Offers best write performance.
l Popular for applications such as storing log files in a database
system.
©Silberschatz, Korth and Sudarshan11.29Database System Concepts 5th Edition
RAID Levels (Cont.)
n RAID Level 2: MemoryStyle ErrorCorrectingCodes (ECC) with bit
striping.
n RAID Level 3: BitInterleaved Parity
l a single parity bit is enough for error correction, not just detection
When writing data, corresponding parity bits must also be
computed and written to a parity bit disk
To recover data in a damaged disk, compute XOR of bits from
other disks (including parity bit disk)
©Silberschatz, Korth and Sudarshan11.30Database System Concepts 5th Edition
RAID Levels (Cont.)
n RAID Level 3 (Cont.)
l Faster data transfer than with a single disk, but fewer I/Os per
second since every disk has to participate in every I/O.
n RAID Level 4: BlockInterleaved Parity; uses blocklevel striping,
and keeps a parity block on a separate disk for corresponding
blocks from N other disks.
l When writing data block, corresponding block of parity bits must
also be computed and written to parity disk
l To find value of a damaged block, compute XOR of bits from
corresponding blocks (including parity block) from other disks.
©Silberschatz, Korth and Sudarshan11.31Database System Concepts 5th Edition
RAID Levels (Cont.)
n RAID Level 4 (Cont.)
l Provides higher I/O rates for independent block reads than Level
3
block read goes to a single disk, so blocks stored on different
disks can be read in parallel
l Before writing a block, parity data must be computed
Can be done by using old parity block, old value of current
block and new value of current block (2 block reads + 2 block
writes)
Or by recomputing the parity value using the new values of
blocks corresponding to the parity block
– More efficient for writing large amounts of data sequentially
l Parity block becomes a bottleneck for independent block writes
since every block write also writes to parity disk
©Silberschatz, Korth and Sudarshan11.32Database System Concepts 5th Edition
RAID Levels (Cont.)
n RAID Level 5: BlockInterleaved Distributed Parity; partitions
data and parity among all N + 1 disks, rather than storing data
in N disks and parity in 1 disk.
l E.g., with 5 disks, parity block for nth set of blocks is
stored on disk (n mod 5) + 1, with the data blocks stored
on the other 4 disks.
©Silberschatz, Korth and Sudarshan11.33Database System Concepts 5th Edition
RAID Levels (Cont.)
n RAID Level 5 (Cont.)
l Higher I/O rates than Level 4.
Block writes occur in parallel if the blocks and their parity
blocks are on different disks.
l Subsumes Level 4: provides same benefits, but avoids
bottleneck of parity disk.
n RAID Level 6: P+Q Redundancy scheme; similar to Level 5, but
stores extra redundant information to guard against multiple disk
failures.
l Better reliability than Level 5 at a higher cost; not used as
widely.
©Silberschatz, Korth and Sudarshan11.34Database System Concepts 5th Edition
Choice of RAID Level
n Factors in choosing RAID level
l Monetary cost
l Performance: Number of I/O operations per second, and
bandwidth during normal operation
l Performance during failure
l Performance during rebuild of failed disk
Including time taken to rebuild failed disk
n RAID 0 is used only when data safety is not important
l E.g. data can be recovered quickly from other sources
n Level 2 and 4 never used since they are subsumed by 3 and 5
n Level 3 is not used since bitstriping forces single block reads to
access all disks, wasting disk arm movement
n Level 6 is rarely used since levels 1 and 5 offer adequate safety
for most applications
n So competition is mainly between 1 and 5
©Silberschatz, Korth and Sudarshan11.35Database System Concepts 5th Edition
Choice of RAID Level (Cont.)
n Level 1 provides much better write performance than level 5
l Level 5 requires at least 2 block reads and 2 block writes to write a
single block, whereas Level 1 only requires 2 block writes
l Level 1 preferred for high update environments such as log disks
n Level 1 had higher storage cost than level 5
l disk drive capacities increasing rapidly (50%/year) whereas disk
access times have decreased much less (x 3 in 10 years)
l I/O requirements have increased greatly, e.g. for Web servers
l When enough disks have been bought to satisfy required rate of I/
O, they often have spare storage capacity
so there is often no extra monetary cost for Level 1!
n Level 5 is preferred for applications with low update rate,
and large amounts of data
n Level 1 is preferred for all other applications
©Silberschatz, Korth and Sudarshan11.36Database System Concepts 5th Edition
Hardware Issues
n Software RAID: RAID implementations done entirely in software,
with no special hardware support
n Hardware RAID: RAID implementations with special hardware
l Use nonvolatile RAM to record writes that are being executed
l Beware: power failure during write can result in corrupted disk
E.g. failure after writing one block but before writing the second
in a mirrored system
Such corrupted data must be detected when power is restored
– Recovery from corruption is similar to recovery from failed
disk
– NVRAM helps to efficiently detected potentially corrupted
blocks
» Otherwise all blocks of disk must be read and compared
with mirror/parity block
©Silberschatz, Korth and Sudarshan11.37Database System Concepts 5th Edition
Hardware Issues (Cont.)
n Hot swapping: replacement of disk while system is running,
without power down
l Supported by some hardware RAID systems,
l reduces time to recovery, and improves availability greatly
n Many systems maintain spare disks which are kept online, and
used as replacements for failed disks immediately on detection
of failure
l Reduces time to recovery greatly
n Many hardware RAID systems ensure that a single point of
failure will not stop the functioning of the system by using
l Redundant power supplies with battery backup
l Multiple controllers and multiple interconnections to guard
against controller/interconnection failures
©Silberschatz, Korth and Sudarshan11.38Database System Concepts 5th Edition
RAID Terminology in the Industry
n RAID terminology not very standard in the industry
l E.g. Many vendors use
RAID 1: for mirroring without striping
RAID 10 or RAID 1+0: for mirroring with striping
l “Hardware RAID” implementations often just offload RAID
processing onto a separate subsystem, but don’t offer
NVRAM.
Read the specs carefully!
n Software RAID supported directly in most operating systems
today
©Silberschatz, Korth and Sudarshan11.39Database System Concepts 5th Edition
Optical Disks
n Compact diskread only memory (CDROM)
l Seek time about 100 msec (optical read head is heavier and
slower)
l Higher latency (3000 RPM) and lower datatransfer rates (36
MB/s) compared to magnetic disks
n Digital Video Disk (DVD)
l DVD5 holds 4.7 GB , variants up to 17 GB
l Slow seek time, for same reasons as CDROM
n Record once versions (CDR and DVDR)
©Silberschatz, Korth and Sudarshan11.40Database System Concepts 5th Edition
Magnetic Tapes
n Hold large volumes of data and provide high transfer rates
l Few GB for DAT (Digital Audio Tape) format, 1040 GB with DLT
(Digital Linear Tape) format, 100 – 400 GB+ with Ultrium format,
and 330 GB with Ampex helical scan format
l Transfer rates from few to 10s of MB/s
n Currently the cheapest storage medium
l Tapes are cheap, but cost of drives is very high
n Very slow access time in comparison to magnetic disks and optical
disks
l limited to sequential access.
l Some formats (Accelis) provide faster seek (10s of seconds) at
cost of lower capacity
n Used mainly for backup, for storage of infrequently used information,
and as an offline medium for transferring information from one system
to another.
n Tape jukeboxes used for very large capacity storage
l (terabyte (1012 bytes) to petabye (1015 bytes)
©Silberschatz, Korth and Sudarshan11.41Database System Concepts 5th Edition
Storage Access
n A database file is partitioned into fixedlength storage units
called blocks. Blocks are units of both storage allocation and
data transfer.
n Database system seeks to minimize the number of block
transfers between the disk and memory. We can reduce the
number of disk accesses by keeping as many blocks as
possible in main memory.
n Buffer – portion of main memory available to store copies of
disk blocks.
n Buffer manager – subsystem responsible for allocating buffer
space in main memory.
©Silberschatz, Korth and Sudarshan11.42Database System Concepts 5th Edition
Buffer Manager
n Programs call on the buffer manager when they need a block from
disk.
n Buffer manager does the following:
l If the block is already in the buffer, return the address of the
block in main memory
2. If the block is not in the buffer
1. Allocate space in the buffer for the block
4 Replacing (throwing out) some other block, if required,
to make space for the new block.
4 Replaced block written back to disk only if it was
modified since the most recent time that it was written
to/fetched from the disk.
4 Read the block from the disk to the buffer, and return the
address of the block in main memory to requester.
©Silberschatz, Korth and Sudarshan11.43Database System Concepts 5th Edition
BufferReplacement Policies
n Most operating systems replace the block least recently used (LRU
strategy)
n Idea behind LRU – use past pattern of block references as a
predictor of future references
n Queries have welldefined access patterns (such as sequential
scans), and a database system can use the information in a user’s
query to predict future references
l LRU can be a bad strategy for certain access patterns involving
repeated scans of data
e.g. when computing the join of 2 relations r and s by a nested loops
for each tuple tr of r do
for each tuple ts of s do
if the tuples tr and ts match
l Mixed strategy with hints on replacement strategy provided
by the query optimizer is preferable
©Silberschatz, Korth and Sudarshan11.44Database System Concepts 5th Edition
BufferReplacement Policies (Cont.)
n Pinned block – memory block that is not allowed to be
written back to disk.
n Tossimmediate strategy – frees the space occupied by a
block as soon as the final tuple of that block has been
processed
n Most recently used (MRU) strategy – system must pin the
block currently being processed. After the final tuple of that
block has been processed, the block is unpinned, and it
becomes the most recently used block.
n Buffer manager can use statistical information regarding the
probability that a request will reference a particular relation
l E.g., the data dictionary is frequently accessed. Heuristic:
keep datadictionary blocks in main memory buffer
n Buffer managers also support forced output of blocks for the
purpose of recovery (more in Chapter 17)
©Silberschatz, Korth and Sudarshan11.45Database System Concepts 5th Edition
File Organization
n The database is stored as a collection of files. Each file is a
sequence of records. A record is a sequence of fields.
n One approach:
lassume record size is fixed
leach file has records of one particular type only
ldifferent files are used for different relations
This case is easiest to implement; will consider variable length
records later.
©Silberschatz, Korth and Sudarshan11.46Database System Concepts 5th Edition
FixedLength Records
n Simple approach:
l Store record i starting from byte n ∗ (i – 1), where n is the
size of each record.
l Record access is simple but records may cross blocks
Modification: do not allow records to cross block
boundaries
n Deletion of record i:
alternatives:
l move records i + 1, . . ., n
to i, . . . , n – 1
l move record n to i
l do not move records, but
link all free records on a
free list
©Silberschatz, Korth and Sudarshan11.47Database System Concepts 5th Edition
Free Lists
n Store the address of the first deleted record in the file header.
n Use this first record to store the address of the second deleted
record, and so on
n Can think of these stored addresses as pointers since they
“point” to the location of a record.
n More space efficient representation: reuse space for normal
attributes of free records to store pointers. (No pointers stored
in inuse records.)
©Silberschatz, Korth and Sudarshan11.48Database System Concepts 5th Edition
VariableLength Records
n Variablelength records arise in database systems in
several ways:
l Storage of multiple record types in a file.
l Record types that allow variable lengths for one or more
fields.
l Record types that allow repeating fields (used in some
older data models).
©Silberschatz, Korth and Sudarshan11.49Database System Concepts 5th Edition
VariableLength Records: Slotted Page
Structure
n Slotted page header contains:
l number of record entries
l end of free space in the block
l location and size of each record
n Records can be moved around within a page to keep them
contiguous with no empty space between them; entry in the
header must be updated.
n Pointers should not point directly to record — instead they
should point to the entry for the record in header.
©Silberschatz, Korth and Sudarshan11.50Database System Concepts 5th Edition
Organization of Records in Files
n Heap – a record can be placed anywhere in the file where
there is space
n Sequential – store records in sequential order, based on the
value of the search key of each record
n Hashing – a hash function computed on some attribute of
each record; the result specifies in which block of the file the
record should be placed
n Records of each relation may be stored in a separate file. In a
multitable clustering file organization records of several
different relations can be stored in the same file
l Motivation: store related records on the same block to
minimize I/O
©Silberschatz, Korth and Sudarshan11.51Database System Concepts 5th Edition
Sequential File Organization
n Suitable for applications that require sequential
processing of the entire file
n The records in the file are ordered by a searchkey
©Silberschatz, Korth and Sudarshan11.52Database System Concepts 5th Edition
Sequential File Organization (Cont.)
n Deletion – use pointer chains
n Insertion –locate the position where the record is to be inserted
l if there is free space insert there
l if no free space, insert the record in an overflow block
l In either case, pointer chain must be updated
n Need to reorganize the file
from time to time to restore
sequential order
©Silberschatz, Korth and Sudarshan11.53Database System Concepts 5th Edition
Multitable Clustering File Organization (cont.)
n Store several relations in one file using a multitable
clustering file organization
n Multitable clustering organization of customer and depositor:
l good for queries involving depositor customer, and for
queries involving one single customer and his accounts
l bad for queries involving only customer
©Silberschatz, Korth and Sudarshan11.54Database System Concepts 5th Edition
Data Dictionary Storage
n Information about relations
l names of relations
l names and types of attributes of each relation
l names and definitions of views
l integrity constraints
n User and accounting information, including passwords
n Statistical and descriptive data
l number of tuples in each relation
n Physical file organization information
l How relation is stored (sequential/hash/)
l Physical location of relation
n Information about indices (Chapter 12)
Data dictionary (also called system catalog) stores
metadata: that is, data about data, such as
©Silberschatz, Korth and Sudarshan11.55Database System Concepts 5th Edition
Data Dictionary Storage (Cont.)
n Catalog structure
l Relational representation on disk
l specialized data structures designed for efficient access, in
memory
n A possible catalog representation:
Relation_metadata = (relation_name, number_of_attributes,
storage_organization, location)
Attribute_metadata = (relation_name, attribute_name,
domain_type,
position, length)
User_metadata = (user_name, encrypted_password, group)
Index_metadata = (relation_name, index_name, index_type,
index_attributes)
View_metadata = (view_name, definition)
Database System Concepts, 5th Ed.
©Silberschatz, Korth and Sudarshan
See www.dbbook.com for conditions on reuse
Extra Slides
©Silberschatz, Korth and Sudarshan11.57Database System Concepts 5th Edition
Record Representation
n Records with fixed length fields are easy to represent
l Similar to records (structs) in programming languages
l Extensions to represent null values
E.g. a bitmap indicating which attributes are null
n Variable length fields can be represented by a pair
(offset, length)
offset: the location within the record, length: field length.
l All fields start at predefined location, but extra indirection
required for variable length fields
Example record structure of account record
account_number
branch_name
balance
PerryridgeA102 40010 000
nullbitmap
Database System Concepts, 5th Ed.
©Silberschatz, Korth and Sudarshan
See www.dbbook.com for conditions on reuse
End of Chapter
©Silberschatz, Korth and Sudarshan11.59Database System Concepts 5th Edition
File Containing account Records
©Silberschatz, Korth and Sudarshan11.60Database System Concepts 5th Edition
File of Figure 11.6, with Record 2 Deleted and
All Records Moved
©Silberschatz, Korth and Sudarshan11.61Database System Concepts 5th Edition
File of Figure 11.6, With Record 2 deleted and
Final Record Moved
©Silberschatz, Korth and Sudarshan11.62Database System Concepts 5th Edition
ByteString Representation of VariableLength
Records
©Silberschatz, Korth and Sudarshan11.63Database System Concepts 5th Edition
Clustering File Structure
©Silberschatz, Korth and Sudarshan11.64Database System Concepts 5th Edition
Clustering File Structure With Pointer Chains
©Silberschatz, Korth and Sudarshan11.65Database System Concepts 5th Edition
The depositor Relation
©Silberschatz, Korth and Sudarshan11.66Database System Concepts 5th Edition
The customer Relation
©Silberschatz, Korth and Sudarshan11.67Database System Concepts 5th Edition
Clustering File Structure
©Silberschatz, Korth and Sudarshan11.68Database System Concepts 5th Edition
©Silberschatz, Korth and Sudarshan11.69Database System Concepts 5th Edition
Figure 11.4
©Silberschatz, Korth and Sudarshan11.70Database System Concepts 5th Edition
Figure 11.7
©Silberschatz, Korth and Sudarshan11.71Database System Concepts 5th Edition
Figure 11.8
©Silberschatz, Korth and Sudarshan11.72Database System Concepts 5th Edition
Figure 11.20
©Silberschatz, Korth and Sudarshan11.73Database System Concepts 5th Edition
ByteString Representation of VariableLength Records
Byte string representation
Attach an endofrecord (⊥) control character to the end of each record
Difficulty with deletion
Difficulty with growth
©Silberschatz, Korth and Sudarshan11.74Database System Concepts 5th Edition
FixedLength Representation
n Use one or more fixed length records:
l reserved space
l pointers
n Reserved space – can use fixedlength records of a known
maximum length; unused space in shorter records filled with a null
or endofrecord symbol.
©Silberschatz, Korth and Sudarshan11.75Database System Concepts 5th Edition
Pointer Method
n Pointer method
l A variablelength record is represented by a list of fixedlength
records, chained together via pointers.
l Can be used even if the maximum record length is not known
©Silberschatz, Korth and Sudarshan11.76Database System Concepts 5th Edition
Pointer Method (Cont.)
n Disadvantage to pointer structure; space is wasted in all
records except the first in a a chain.
n Solution is to allow two kinds of block in file:
l Anchor block – contains the first records of chain
l Overflow block – contains records other than those that
are the first records of chairs.
©Silberschatz, Korth and Sudarshan11.77Database System Concepts 5th Edition
Mapping of Objects to Files
n Mapping objects to files is similar to mapping tuples to files in a relational
system; object data can be stored using file structures.
n Objects in OO databases may lack uniformity and may be very large;
such objects have to managed differently from records in a relational
system.
l Set fields with a small number of elements may be implemented
using data structures such as linked lists.
l Set fields with a larger number of elements may be implemented as
separate relations in the database.
l Set fields can also be eliminated at the storage level by
normalization.
Similar to conversion of multivalued attributes of ER diagrams to
relations
©Silberschatz, Korth and Sudarshan11.78Database System Concepts 5th Edition
Mapping of Objects to Files (Cont.)
n Objects are identified by an object identifier (OID); the storage system
needs a mechanism to locate an object given its OID (this action is
called dereferencing).
l logical identifiers do not directly specify an object’s physical
location; must maintain an index that maps an OID to the object’s
actual location.
l physical identifiers encode the location of the object so the
object can be found directly. Physical OIDs typically have the
following parts:
1. a volume or file identifier
2. a page identifier within the volume or file
3. an offset within the page
©Silberschatz, Korth and Sudarshan11.79Database System Concepts 5th Edition
Management of Persistent Pointers
n Physical OIDs may be a unique identifier. This identifier is
stored in the object also and is used to detect references via
dangling pointers.
©Silberschatz, Korth and Sudarshan11.80Database System Concepts 5th Edition
Management of Persistent Pointers
(Cont.)
n Implement persistent pointers using OIDs; persistent pointers are
substantially longer than are inmemory pointers
n Pointer swizzling cuts down on cost of locating persistent objects
already inmemory.
n Software swizzling (swizzling on pointer deference)
l When a persistent pointer is first dereferenced, the pointer is
swizzled (replaced by an inmemory pointer) after the object is
located in memory.
l Subsequent dereferences of of the same pointer become cheap.
l The physical location of an object in memory must not change if
swizzled pointers pont to it; the solution is to pin pages in memory
l When an object is written back to disk, any swizzled pointers it
contains need to be unswizzled.
©Silberschatz, Korth and Sudarshan11.81Database System Concepts 5th Edition
Hardware Swizzling
n With hardware swizzling, persistent pointers in objects need the
same amount of space as inmemory pointers — extra storage
external to the object is used to store rest of pointer information.
n Uses virtual memory translation mechanism to efficiently and
transparently convert between persistent pointers and inmemory
pointers.
n All persistent pointers in a page are swizzled when the page is
first read in.
l thus programmers have to work with just one type of pointer,
i.e., inmemory pointer.
l some of the swizzled pointers may point to virtual memory
addresses that are currently not allocated any real memory
(and do not contain valid data)
©Silberschatz, Korth and Sudarshan11.82Database System Concepts 5th Edition
Hardware Swizzling
n Persistent pointer is conceptually split into two parts: a page identifier,
and an offset within the page.
l The page identifier in a pointer is a short indirect pointer: Each
page has a translation table that provides a mapping from the
short page identifiers to full database page identifiers.
l Translation table for a page is small (at most 1024 pointers in a
4096 byte page with 4 byte pointer)
l Multiple pointers in page to the same page share same entry in
the translation table.
©Silberschatz, Korth and Sudarshan11.83Database System Concepts 5th Edition
Hardware Swizzling (Cont.)
n Page image before swizzling (page located on disk)
©Silberschatz, Korth and Sudarshan11.84Database System Concepts 5th Edition
Hardware Swizzling (Cont.)
n When system loads a page into memory the persistent pointers in the page
are swizzled as described below
1. Persistent pointers in each object in the page are located using object
type information
n For each persistent pointer (pi, oi) find its full page ID Pi
H If Pi does not already have a virtual memory page allocated to it,
allocate a virtual memory page to Pi and readprotect the page
Note: there need not be any physical space (whether in memory
or on disk swapspace) allocated for the virtual memory page at
this point. Space can be allocated later if (and when) Pi is
accessed. In this case readprotection is not required.
Accessing a memory location in the page in the will result in a
segmentation violation, which is handled as described later
H Let vi be the virtual page allocated to Pi (either earlier or above)
H Replace (pi, oi) by (vi, oi)
H Replace each entry (pi, Pi) in the translation table, by (vi, Pi)
©Silberschatz, Korth and Sudarshan11.85Database System Concepts 5th Edition
Hardware Swizzling (Cont.)
n When an inmemory pointer is dereferenced, if the operating
system detects the page it points to has not yet been allocated
storage, or is readprotected, a segmentation violation occurs.
n The mmap() call in Unix is used to specify a function to be invoked
on segmentation violation
n The function does the following when it is invoked
1. Allocate storage (swapspace) for the page containing the
referenced address, if storage has not been allocated earlier.
Turn off readprotection
2. Read in the page from disk
3. Perform pointer swizzling for each persistent pointer in the
page, as described earlier
©Silberschatz, Korth and Sudarshan11.86Database System Concepts 5th Edition
Hardware Swizzling (Cont.)
n Page with short page identifier 2395 was allocated address 5001.
Observe change in pointers and translation table.
n Page with short page identifier 4867 has been allocated address
4867. No change in pointer and translation table.
Page image after swizzling
©Silberschatz, Korth and Sudarshan11.87Database System Concepts 5th Edition
Hardware Swizzling (Cont.)
n After swizzling, all short page identifiers point to virtual memory addresses
allocated for the corresponding pages
l functions accessing the objects are not even aware that it has
persistent pointers, and do not need to be changed in any way!
l can reuse existing code and libraries that use inmemory pointers
n After this, the pointer dereference that triggered the swizzling can continue
n Optimizations:
l If all pages are allocated the same address as in the short page
identifier, no changes required in the page!
l No need for deswizzling — swizzled page can be saved asis to disk
l A set of pages (segment) can share one translation table. Pages can
still be swizzled as and when fetched (old copy of translation table is
needed).
n A process should not access more pages than size of virtual memory —
reuse of virtual memory addresses for other pages is expensive
©Silberschatz, Korth and Sudarshan11.88Database System Concepts 5th Edition
Disk versus Memory Structure of Objects
n The format in which objects are stored in memory may be different from
the formal in which they are stored on disk in the database. Reasons
are:
l software swizzling – structure of persistent and inmemory pointers
are different
l database accessible from different machines, with different data
representations
l Make the physical representation of objects in the database
independent of the machine and the compiler.
l Can transparently convert from disk representation to form required
on the specific machine, language, and compiler, when the object
(or page) is brought into memory.
©Silberschatz, Korth and Sudarshan11.89Database System Concepts 5th Edition
Large Objects
n Large objects : binary large objects (blobs) and character large
objects (clobs)
l Examples include:
text documents
graphical data such as images and computer aided designs
audio and video data
n Large objects may need to be stored in a contiguous sequence of
bytes when brought into memory.
l If an object is bigger than a page, contiguous pages of the buffer
pool must be allocated to store it.
l May be preferable to disallow direct access to data, and only allow
access through a filesystemlike API, to remove need for
contiguous storage.
©Silberschatz, Korth and Sudarshan11.90Database System Concepts 5th Edition
Modifying Large Objects
n If the application requires insert/delete of bytes from specified regions of
an object:
l B+tree file organization (described later in Chapter 12) can be
modified to represent large objects
l Each leaf page of the tree stores between half and 1 page worth of
data from the object
n Specialpurpose application programs outside the database are used to
manipulate large objects:
l Text data treated as a byte string manipulated by editors and
formatters.
l Graphical data and audio/video data is typically created and displayed
by separate application
l checkout/checkin method for concurrency control and creation of
versions
Các file đính kèm theo tài liệu này:
- ch11_1039_3224.pdf