tips for managing a vldb

15
Managing VLDBs Options to ease maintenance

Upload: john-martin

Post on 10-Apr-2017

88 views

Category:

Technology


0 download

TRANSCRIPT

Page 1: Tips for managing a VLDB

Managing VLDBs

Options to ease maintenance

Page 2: Tips for managing a VLDB

About Me

John Q MartinSales Engineer for SQL Sentry

Over a decade of experience with SQL Server as a Dev, DBA and BI Dev. Worked as a Premier Field Engineer with Microsoft UK prior to joining SQL Sentry.

Twitter : @SQLDiplomatEmail : [email protected] : http://blogs.sqlsentry.com/author/JohnMartin/

Page 3: Tips for managing a VLDB

Book a Demo with me:http://SQLSentry.com/BookJohn

Page 4: Tips for managing a VLDB

http://www.sqlsentry.com/sql-server-books/

Free Resources

Page 5: Tips for managing a VLDB

What is a VLDB? Very Large Database The context of what constitutes Very Large will vary,

depending on experience and what you are geared up to work with. Ranging from Gigabytes to Terabytes.

Not to be confused with ‘Big Data’ Related to the Volume, Variety, Velocity, and Veracity of data

being processed.

Page 6: Tips for managing a VLDB

Typical Management Issues

Disaster Recovery Planning and Implementation

Index and Statistics maintenance

Consistency checking

Performance Tuning

Page 7: Tips for managing a VLDB

Disaster Recovery Planning Clearly defined Recovery Time Objective (RTO) and

Recovery Point Objective (RPO) requirements are vital. No matter how well you tune a system, you will always run

into a bottleneck

Constantly reassess and test your plans around restore capabilities for a VLDB. Backup performance will be monitored, restore performance is

not

Page 8: Tips for managing a VLDB

Disaster Recovery Planning There are a number of options available to you when it

comes to improving the performance of backups; Enable Compression Stripe Backup Files Isolate backup storage paths Modify BUFFERCOUNT, MAXTRANSFERSIZE and BLOCKSIZE

options for the BACKUP statement

Page 9: Tips for managing a VLDB

Disaster Recovery Planning SAN Snapshots are a double-edged sword;

Yes : Snapshots are fast for large databases But : You are exposed while the snapshot is only on the SAN. But : You cannot perform a point in time restore, you can only

restore to a point in time…

Page 10: Tips for managing a VLDB

Disaster Recovery Planning SQL Server has additional backup options to help speed

things up. Differential File Group Backup File Backup

When using Enterprise Edition, piecemeal restores mean that databases can be partially online. Architect your databases appropriately

Page 11: Tips for managing a VLDB

Index & Stats Maintenance Rebuilding large indexes can be prohibitive. From space

requirements for Online operations, to the duration taken for offline operations.

Options for managing Index maintenance; Table Partitioning Online Index rebuilds Reorganize rather than Rebuild

Page 12: Tips for managing a VLDB

Index & Stats Maintenance Managing Statistics in VLDB scenarios can be

problematic for a number of reasons; Statistics Skew Auto-update threshold is disproportionate Size of tables means rebuilding with FULLSCAN is not an option

Options for managing statistics Trace Flag 2371 Table Partitioning & Incremental Statistics (SQL 2014)

Page 13: Tips for managing a VLDB

CHECKDB CHECKDB, while not a blocking process can saturate the IO sub-system. For large databases this can also use a lot of TempDB resources, and impact production workloads.

Breaking CHECKDB down into is constituent parts allows you to nibble through a consistency check. CHECKALLOC CHECKCATALOG CHECKTABLE

Page 14: Tips for managing a VLDB

Performance Tuning Understanding your workload is key to unlocking

performance; Are there any good candidates for table partitioning? What are the costs for the plans in cache Is your database schema correct

Leverage tracing and RML utilities to identify queries that are being issued to the database.

Page 15: Tips for managing a VLDB