未加星标

TSQL Tuesday #83 The Stats update solution

字体大小 | |
[数据库(mssql) 所属分类 数据库(mssql) | 发布者 店小二05 | 时间 2016 | 作者 红领巾 ] 0人收藏点击收藏

TSQL Tuesday #83   The Stats update solution

TSQL Tuesday is a monthly blog part hosted by a different blogger every month it was started by Adam Machanic . This week’s TSQL Tuesday is hosted by Andy Mallon the topic is ‘We’re dealing with the same problem’. I have chosen to write about a common problem I have encountered at many places I have worked at. The problem has to do with queries or stored procedures that suddenly start to perform poorly when no changes have been made to the code or to the server.

The common perception/misunderstanding I have encountered for this is that it is only an issue with statistics updates and updating statistics with full scan should take care of it. In many cases this is the real reason. In some cases it really isn’t. It could be an issue with parameter sniffing, and a plan being reused that was generated for a set of parameters and appropriate for that set of parameters. But, most people jump to fixing statistics. This is especially true when they don’t really have the runtime plan it used, can’t find it in the cache and are just going with some sort of past experience.

At one place I was at people would update statistics every 10 minutes or so in a frantic attempt to ‘fix the slow query’, which would at many times not respond at all. At another place they actually had an automated check for when the stored procedure finished running and if it was still running beyond its normal duration a full scan statistics update would fire off. None of these are wrong but repeatedly doing the same thing when the query does not improve, and assuming that is the only reason for the problem is wrong.

What I do is to recompile the plan with different parameters test it to see the performance. If you are able to get a difference in performance with a different set of parameters then it is probably a parameter sniffing issue. And, if the stats updates do not fix it it is probably that too. Statistics updates are never really a bad thing to do, butthey may not fix every slow query there is. Check if the issue is with parameter sniffing also. Also make sure there are no changes that went out to the code and to the environment, that may be contributing to it.


TSQL Tuesday #83   The Stats update solution

本文数据库(mssql)相关术语:熊片数据库 mssql数据库 oracle数据库 pubmed数据库 access数据库 万方数据库

主题: SQL
分页:12
转载请注明
本文标题:TSQL Tuesday #83 The Stats update solution
本站链接:http://www.codesec.net/view/482277.html
分享请点击:


1.凡CodeSecTeam转载的文章,均出自其它媒体或其他官网介绍,目的在于传递更多的信息,并不代表本站赞同其观点和其真实性负责;
2.转载的文章仅代表原创作者观点,与本站无关。其原创性以及文中陈述文字和内容未经本站证实,本站对该文以及其中全部或者部分内容、文字的真实性、完整性、及时性,不作出任何保证或承若;
3.如本站转载稿涉及版权等问题,请作者及时联系本站,我们会及时处理。
登录后可拥有收藏文章、关注作者等权限...
技术大类 技术大类 | 数据库(mssql) | 评论(0) | 阅读(40)