Basics Storing Data on Disks and Files Chapter 9 1
Disks and Files v DBMS stores information on (“hard”) disks. v This has major implications for DBMS design! § § § READ: transfer data from disk to main memory (RAM). WRITE: transfer data from RAM to disk. Both are high-cost operations, relative to in-memory operations, so must be planned carefully! 2
Why Not Store Everything in Main Memory? v Costs too much. § Same amount of money will buy you say either 128 MB of RAM or 20 GB of disk. v Main memory is volatile. § We want data to be saved between runs. (Obviously!) v Typical storage hierarchies: § § § Main memory (RAM) for currently used data (primary storage). Disk for the main database (secondary storage). Tapes for archiving older versions of data (tertiary storage). 3
Disks Secondary storage device of choice. v Main advantage over tapes: v § random access vs. sequential. v Data is stored and retrieved in units : § called disk blocks or pages. v Unlike RAM, time to retrieve a disk page varies depending upon location on disk. § Therefore, relative placement of pages on disk has major impact on DBMS performance! 4
Components of a Disk Spindle Tracks Disk head The platters spin (say, 90 rps). Sector v. The arm assembly is moved in or out to position a head on a desired track. v. Tracks under heads Platters Arm movement make a cylinder (imaginary!). v. Only one head reads/writes at any Arm assembly one time. v Block size is a multiple of sector size (which is fixed). v 5
Accessing a Disk Page v Time to access (read/write) a disk block: § § § v Seek time and rotational delay dominate. § § § v seek time (moving arms to position disk head on track) rotational delay (waiting for block to rotate under head) transfer time (actually moving data to/from disk surface) Seek time varies from about 1 to 20 msec Rotational delay varies from 0 to 10 msec Transfer rate is about 1 msec per 4 KB page Lower I/O cost: reduce seek/rotation delays! 6
Arranging Pages on Disk v `Next’ block concept: § blocks on same track, followed by § blocks on same cylinder, followed by § blocks on adjacent cylinder v Blocks in a file should be arranged sequentially on disk (by `next’), to minimize seek and rotational delay. v For a sequential scan, pre-fetching several pages at a time is a big win! 7
Disk Space Management v Lowest layer of DBMS software manages space on disk. v Higher levels call upon this layer to: § § v allocate/de-allocate a page read/write a page Higher levels don’t need to know how this is done, or how free space is managed. 13
Buffer Management in a DBMS Page Requests from Higher Levels BUFFER POOL disk page free frame MAIN MEMORY DISK DB choice of frame dictated by replacement policy Data must be in RAM for DBMS to operate on it! v Table of <frame#, pageid> pairs is maintained. v 14
When a Page is Requested. . . v If requested page is not in buffer pool: § § § v Choose a frame for replacement If frame is dirty, write it to disk Read requested page into chosen frame Pin the page and return its address. If requests can be predicted (e. g. , sequential scans) pages can be pre-fetched (several pages at a time)! * 15
More on Buffer Management v Requestor of page must unpin it, and indicate whether page has been modified: § v Page in pool may be requested many times, § § v dirty bit is used for this. a pin count is used. A page is a candidate for replacement iff pin count = 0. CC & recovery may entail additional I/O when a frame is chosen for replacement. (Write-Ahead Log protocol; more later. ) 16
DBMS vs. OS File System OS does disk space & buffer management already! So why not let OS manage these tasks? v v v Differences in OS support: Portability issues Some limitations, e. g. , files don’t span multiple disk devices. Buffer management in DBMS requires ability to: § pin a page in buffer pool, § force a page to disk (important for CC & recovery), § adjust replacement policy, and pre-fetch pages based on access patterns in typical DB operations. 18
Structure of a DBMS v v v A typical DBMS has a layered architecture. Disk Storage hierarchy, RAID Disk Space Management Roles, Free blocks v Buffer Management Buffer Pool, Replacement policy v Files and Access Methods File organization (heap files, sorted file, indexes) File and page level storage (collection of pages or records) These layers must consider concurrency control and recovery Query Optimization and Execution Relational Operators Files and Access Methods Buffer Management Disk Space Management Index Files Data Files DBSystem Catalog 19
Files of Records Page or block is the granularity for doing I/O v Higher levels of DBMS operate on : v v § records, and § files composed of records. FILE: A collection of pages, each containing a collection of records. v File must support: § § § insert/delete/modify record read a particular record (specified using record id) scan all records (possibly with some conditions on the records to be retrieved) 20
Unordered Files (Heap Files) v Simplest file structure contains records in no particular order. v As file grows and shrinks, disk pages are allocated and de-allocated. v To support record level operations, we must: § § § v keep track of the pages in a file keep track of free space on pages keep track of the records on a page There are many alternatives for keeping track of this. 21
Alternative 1: Heap File Implemented as List Data Page Full Pages Header Page Data Page v v Data Pages with Free Space Maintain a table containing pairs of: <heap_file_name, head_page_address> Each page contains 2 `pointers’ (rid) plus data. 22
Heap File Implemented as a List v Insert a new page into heap file § Disk manager adds a new free space page into link v Delete a page from heap file § Removed from the list § Disk manager deallocates it v Disadvantages: § If records are of variable length, all pages will be in free list. § Retrieve and examine several pages for enough space. 23
Alternative 2: Heap File Using Page Directory Data Page 1 Header Page Data Page 2 DIRECTORY v v Data Page N In directory, each entry for a page includes number of free bytes on page. The directory is a collection of pages (linked list implementation is just one alternative). § Much smaller than linked list of all HF pages! 24
Alternative 2: Heap File Using a Page Directory v Advantage of Page Directory : § The size of directory is very small (much smaller than heap file. ) § Searching space is very efficient, because find free space without looking at actual heap data pages. 25
Page Formats v Page : abstraction is used for I/O Record : data granularity for higher level of DBMS v How to arrange records in pages? v § Identify a record: • <page_id, slot_number>, where slot_number = rid • Most cases, use <page_id, slot_number> as rid. v v Alternative approaches to manage slots on a page How to support insert/deleting/searching? 26
Records Formats: Fixed Length Record F 1 F 2 F 3 F 4 L 1 L 2 L 3 L 4 Base address (B) Address = B+L 1+L 2 Information about field types same for all records in a file v Stored record format in system catalogs. + Finding i’th field does not require scan of record, just offset calculation. v 27
Page Formats: Fixed Length Records Slot 1 Slot 2 Free Space . . . Slot N Slot M N PACKED 1. . . 0 1 1 M number of records M. . . 3 2 1 UNPACKED, BITMAP number of slots * Record id = <page id, slot #>. * Note: In first alternative, moving records for free space management changes rid; may not be acceptable if existing external references to the record that is moved. 28
Record Formats: Variable Length v Two alternative formats (# fields is fixed): F 1 4 Field Count F 2 $ F 3 $ F 4 $ $ Fields Delimited by Special Symbols F 1 F 2 F 3 F 4 Array of Field Offsets + Second offers direct access to i’th field + efficient storage of nulls ; - small directory overhead. 29
Page Formats: Variable Length Records Rid = (i, N) Length = 20 Page i Rid = (i, 2) Length = 16 Rid = (i, 1) 20 N . . . 16 2 Length = 24 N 24 1 # slots SLOT DIRECTORY * Offset of record from start of data area Pointer to start of free space Slot directory = {<record_offset, record_length>} 30
Page Formats: Variable Length Records Slot directory = {<record_offset, record_length>} * Dis/Advantages: + Moving: rid is not changed + Deletion: offset = -1 (rid changed? Can we delete slot? Why? ) + Insertion: Reuse deleted slot. Only insert if none available. * * Free space? Free space pointer? Recycle after deletion? 31
System Catalogs v Meta information stored in system catalogs. v For each index: § v For each relation: § § v name, file structure (e. g. , Heap file) attribute name and type, for each attribute index name, for each index integrity constraints For each view: § v structure (e. g. , B+ tree) and search key fields view name and definition Plus statistics, authorization, buffer pool size, etc. * Catalogs are themselves stored as relations! 32
Attr_Cat(attr_name, rel_name, type, position) 33
Summary v Disks provide cheap, non-volatile storage. § § v Random access, but cost depends on location of page on disk Important to arrange data sequentially to minimize seek and rotation delays. Buffer manager brings pages into RAM. § § Page stays in RAM until released by requestor. Written to disk when frame chosen for replacement. Frame to replace based on replacement policy. Tries to pre-fetch several pages at a time. 34
Скачать презентацию Basics Storing Data on Disks and Files Chapter
135d9703c4acf42b04509c6a52a9eefc.ppt