SQLServer Database Structures
1 like846 views
The document provides details about an SQL expert's background and certifications. It summarizes the expert's career starting in 1982 working with computers and 1988 starting in the computer industry. In 1996, they started working with SQL Server 6.0 and have since earned multiple Microsoft certifications. The expert now provides training and consultation services, and created an online school called SQL School Greece to teach SQL Server.
![DBCC PAGE
DBCC PAGE ({dbid | dbname}, filenum, pagenum[, printopt])
DBCC TRACEON(3604);
GO
DBCC PAGE (AdventureWorks2012, 1, 266, 1);](https://image.slidesharecdn.com/sqlsaturrdayathens2018-sqlserverdatabasestructures-180621214421/85/SQLServer-Database-Structures-18-320.jpg)
SQLServer Database Structures
- 2. PRESENTERINFO 1982 I started working with computers 1988 I started my professional career in computers industry 1996 I started working with SQL Server 6.0 1998 I earned my first certification at Microsoft as Microsoft Certified Solution Developer (3rd in Greece) 1999 I started my career as Microsoft Certified Trainer (MCT) with more than 30.000 hours of training until now! 2010 I became for first time Microsoft MVP on Data Platform I created the SQL School Greece www.sqlschool.gr 2012 I became MCT Regional Lead by Microsoft Learning Program. 2013 I was certified as MCSE : Data Platform I was certified as MCSE : Business Intelligence 2016 I was certified as MCSE: Data Management & Analytics 2017 Certified as MCSA : Machine Learning Recertified as MCSE: Data Management & Analytics Antonios Chatzipavlis Data Solutions Consultant & Trainer MVP on Data Platform MCT, MCSE, MCITP, MCPD, MCSD, MCDBA, MCSA, MCTS, MCAD, MCP, OCA, ITIL-F
- 3. Μια πηγή ενημέρωσης για την Microsoft Data Platform προς τους Έλληνες IT Professionals, DBAs, Developers, Information Workers αλλά και απλούς χομπίστες που απλά τους αρέσει ο SQL Server. Help line : help@sqlschool.gr • Articles about SQL Server • SQL Server News • SQL Nights • Webcasts • Downloads • Resources What we are doing here Follow us in socials fb/sqlschoolgr fb/groups/sqlschool @antoniosch @sqlschool yt/c/SqlschoolGr SQL School Greece group SELECT KNOWLEDGE FROM SQL SERVER
- 4. Connect with PASS Sign up for a free membership today at: pass.org is the PASS Local Group in Greece
- 6. SPONSORS
- 7. PRESENTATIONCONTENT • Database Structure Internals • Data File Internals • tempdb Internals
- 9. • 32,767 Databases per instance • 524,767 TB is the max size per database • 16 TB is the max size for each data file • 2 TB is the max size for each log file • 32,767 files per database • 32,767 filegroups per database • 64 KB is the size of each Extent - 8 pages X 8 KB per page • 8 KB is the size of each Page DATABASE COMPONENTS Database File Group (one or more) Data File (one or more) Extent (mixed or uniform) Pages (8 KB each)
- 10. File groups • Logical grouping of data files • Can bring administration and performance benefits • FILESTREAM has its own file group FILEGROUPS AND DATABASE FILES Database Files • Data Files - At least one per database - Multiple files can: - improve performance, - aid maintainability - circumvent operating system file size limitations • Transaction Log Files - Usually only one per database - No performance advantage from multiple files
- 11. EXTENTS Comprises eight physically continuous pages Mixed shared between different database objects Uniform owned by a single database object
- 12. PAGE STRUCTURE AND PAGE TYPES Page Types • Data (1) • Index (2) • Text (3 and 4) • Boot (13) • File Header (15) • PFS (11) • GAM (8) • SGAM (9) • IAM (10) • DIFF_MAP (16) • ML_MAP (17)
- 13. PAGE HEADER pageID The file number and page number of this page in the database Metadata: ObjectId The ID of the object to which this page belongs slotCnt The total number of slots (rows) used on this page Metadata: PartitionId The ID of the partition that this page is part of Type Page Type Metadata: AllocUnitId The ID of the allocation unit that contains this page LSN The LSN corresponding to the last log entry that changed this page freeData The byte offset of the first free space on this page nextPage The file number and page number of the next page if this page is in a page chain Pminlen The number of bytes in fixed-length portion of rows prevPage The file number and page number of the previous page if this page is in a page chain freeCnt The number of free bytes on the page Level The level of this page in an index (always 0 for leaf pages) indexId The index ID of this page (always 0 for data pages)
- 14. • Data rows for in-row data - The maximum size of a single data row is 8,060 bytes of in-row data. - Rows can also have row-overflow and LOB data stored on separate pages. - The number of rows stored on a given page varies depending on the table structure and on the data being stored. - A table with all fixed-length columns always can store the same number of rows per page - Variable-length rows can store as many rows as will fit based on the actual length of the data entered. - Keeping the row length shorter allows more rows to fit on a page, thus reducing I/O and increasing the likelihood that needed data can stay in cache. • Row offset array - The row offset array is a block of 2-byte entries - Each indicating the offset on the page at which the corresponding data row begins. - The row offset array indicates the logical order of rows on a page. - For example, if a table has a clustered index, SQL Server stores the rows in the order of the clustered index key. - This doesn’t mean the rows are physically stored on the page in the order of the clustered index key. - Rather, slot 0 in the offset array refers to the first row in the clustered index key order, slot 1 refers to the second row, and so forth DATA ROWS AND ROW OFFSET ARRAY
- 15. • Status Bits A - Bit 0. This indicates special versioning information. In SQL Server 2012, this is always 0. - Bits 1 through 3. Taken as a 3-bit value, - 0(000) indicates a primary record, - 1 (001) indicates a forwarded record, - 2 (010) indicates a forwarding stub, - 3 (011) indicates an index record, - 4 (100) indicates a BLOB fragment or row-overflow data - 5(101) indicates a ghost index record - 6(110) indicates a ghost data record, - 7(111) indicates a ghost version record. - Bit 4. This indicates that a NULL bitmap exists. In SQL Server 2012, a NULL bitmap is always present, even if no NULLs are allowed in any column. - Bit 5. This indicates that variable-length columns exist in the row. - Bit 6. This indicates that the row contains versioning information. - Bit 7. This isn’t used in SQL Server 2012. • Status Bits B - The only bit used in the Status Bits B field indicates that the record is a ghost forwarded record. RECORD STRUCTURE AND RECORD TYPES
- 16. • File Header Page (FH) • Page Free Space (PFS) • Global Allocation Map (GAM) • Shared Global Allocation Map (SGAM) • Differential Change Map (DCM) • Bulk Change Map • Boot Page • Index Allocation Map (IAM) ALLOCATION BITMAPS AND SPECIAL PAGES
- 17. • Page allocation •The process of allocating new pages to an object • Allocation units •IN_ROW_DATA •LOB_DATA •ROW_OVERFLOW_DATA • Page allocations can be analysed using DBCC page PAGE ALLOCATION AND ALLOCATION UNITS
- 18. DBCC PAGE DBCC PAGE ({dbid | dbname}, filenum, pagenum[, printopt]) DBCC TRACEON(3604); GO DBCC PAGE (AdventureWorks2012, 1, 266, 1);
- 21. VOLUME CONFIGURATION BEST PRACTICES Volume RAID 10 for transaction log files RAID 10 or RAID 5 for database data files Store data and transaction log files on physically separate volumes Defragment disks frequently Size data and transaction log files to avoid auto growth
- 22. NUMBER OF DATA FILES AND PL ACEMENT For most databases a single data file is sufficient: To improve performance:
- 23. • Logical I/O - pages read from cache • Physical I/O - pages read from disk • You can view physical and logical I/O stats using - the DMV sys.dm_exec_query_stats or - by using the Transact-SQL command SET STATISTICS IO ON before issuing a query. - SET STATISTICS IO ON will only return statistics for queries in the current session. PHYSICAL I/O VS. LOGICAL I/O • Buffer cache hit ratio - percentage of pages read from cache without having to be read from disk - calculated by the formula ((logical reads–physical reads)/logical reads) * 100. - a SQL Server instance that is performing well will have a buffer cache hit ratio close to 100. - The buffer cache hit ratio can be improved by making more memory available to SQL Server.
- 24. ALLOCATIONS WITHIN DATA FILES Allocation is simple in a file group that has a single file In a multifile group, SQL Server uses: • Round robin allocation • Proportional fill
- 25. INSTANT FILE INITIALIZATION Improves performance by skipping zeroing of data pages on file creation and growth Best practice is to enable following assessment of security risk Disabled by default: security risk
- 26. Auto Grow • Automatically expands a file when space is low • Manually manage file size for better performance • Auto grow best practices: - Leave enabled to avoid downtime in case file size limit is reached - Set to a fixed rather than a percentage AUTO GROW AND AUTO SHRINK Auto Shrink • Shrinks data file automatically to release disk space • Use with caution as: - Auto shrink causes index fragmentation - It’s uncontrollable and affects performance - It can lead to growth-shrink cycle severely affecting performance
- 27. DATA FILE SHRINKING Shrink • Causes index fragmentation • Resource intensive • Degrades performance Best used for • Emptying a file before removing it • Changing a file or file group to read only • Reclaiming free space following large delete operation Alternative • Move all indexes to a new file group • Move heaps with shrink. Heaps are not fragmented with shrink • Remove old file group
- 28. Database file configuration • SSMS database properties • sp_helpfile and sp_helpfilegroup • sys.database_files and sys.filegroups • sys.master_files • sys.dm_db_file_space_usage MONITORING DATABASE FILES Database file activity •SSMS Activity Monitor – Data File I/O •Wait statistics – PAGEIOLATCH_*, WRITELOG •sys.dm_io_virtual_file_stats
- 30. TEMPDB INTERNALS
- 31. TEMPDB USAGE tempdb tempdb Stores: • Internal objects • Version store • Some user objects Recreated at startup Must not be allowed to run out of space Should be actively
- 33. VERSION STORE INTERNALS Stores all version records from all databases New allocation unit created every minute Cleanup happens every minute Nonlogged operation Version Store
- 34. Configuration • Enable instant file initialization • Size for expected workload • Isolate tempdb data files • Use multiple data files • Enable auto growth as a contingency TEMPDB CONFIGURATION Prohibited operations • Backup and restore • Add file groups • Change collation • Set offline or read only tempdb size can be monitored using the sys.dm_db_file_space_usage DMV
- 37. EVALUATIONS http://bit.ly/2xUr3XP EVENT EVALUATION LINK SESSIONS EVALUATION LINK http://bit.ly/2snEVoq
- 38. Thank you
- 39. SELECT KNOWLEDGE FROM SQL SERVER Copyright © 2018 SQLschool.gr. All right reserved. PRESENTER MAKES NO WARRANTIES, EXPRESS, IMPLIED OR STATUTORY, AS TO THE INFORMATION IN THIS PRESENTATION