Submitted by yejr on 周一, 2008/08/18 - 07:13
现场气氛就是好吖,中国观众真是热情,日本人特别被安排在我们后方不远的地方,不过加油声音远比我们小多了。最后女排不负众望,3:0完美拿下。开局差点大比分落后,还好追上来了,没让我的手掌白白拍红了,欧耶。
Submitted by yejr on 周四, 2008/08/07 - 13:59
原文出自: http://www.mysqlperformanceblog.com/2008/07/31/percona-rpms-for-redhat-5-centos-5-x86_64/
经过测试,和官方的发行版相当,性能略有下降。但是新增了microsecond slow query log和innodb index statistic对很多人来说都很有帮助,还是值得的。有兴趣的朋友可以试试看。
Submitted by yejr on 周六, 2008/07/26 - 22:55
XFS 简介
XFS 最初是由 Silicon Graphics,Inc. 于 90 年代初开发的。那时,SGI 发现他们的现有文件系统(existing
filesystem,EFS)正在迅速变得不适应当时激烈的计算竞争。为解决这个问题,SGI 决定设计一种全新的高性能 64
位文件系统,而不是试图调整 EFS在先天设计上的某些缺陷。因此,XFS 诞生了,并于 1994 年随 IRIX 5.3
的发布而应用于计算。它至今仍作为 SGI 基于 IRIX 的产品(从工作站到超级计算机)的底层文件系统来使用。现在,XFS 也可以用于
Linux。XFS 的 Linux 版的到来是激动人心的,首先因为它为 Linux
社区提供了一种健壮的、优秀的以及功能丰富的文件系统,并且这种文件系统所具有的可伸缩性能够满足最苛刻的存储需求。
上面介绍转自:http://www.ibm.com/developerworks/cn/linux/filesystem/l-fs9/index.html,更多关于其他文件系统的介绍请看ibm的系列文章:文件系统系列。
Submitted by yejr on 周五, 2008/07/25 - 15:43
1、前言
PBXT 是 PrimeBase 公司推出的MySQL插件引擎,其功能和 InnoDB 类似,主要特性如下:
- MVCC Support:
MVCC stands for Multi-version Concurrency Control. MVCC allows reading the database without locking.
- Fully ACID complient:
This means that transactionally safe, and able to handle multiple concurrent transactions.
- Row-level locking:
When updating, PBXT uses row-level locking. Row-level locking is also used during SELECT FOR UPDATE.
- Fast Rollback and Recovery:
PBXT
uses a specialized method to identify garbage which makes "undo"
unncessary. This make both rollback of transactions and recovery after
restart very fast.
- Deadlock Detection:
PBXT identifies all kinds of deadlocks immediately.
- Write-once:
PBXT
uses a log-based storage which makes it possible to write transactional
data directly to the database, without first being writen to the
transaction log.
- Referential Integrity:
PBXT supports foreign key definitions, including cascaded updates and deletes.
- BLOB streaming:
In combination with the BLOB Streaming engine PBXT can stream binary and media directly in and out of the database.
Submitted by yejr on 周四, 2008/07/24 - 09:39
原文来自 http://openquery.com.au/products/about-drizzle
About Drizzle
Drizzle is described as a "Lightweight SQL Database for Cloud and Web".
The project is about building a database designed for the typical Web
usage patterns for a database. It is being designed for massive
concurrency on modern multi-cpu/core architecture. The code is
originally derived from MySQL.
Submitted by yejr on 周三, 2008/07/23 - 15:22
声明:本文拒绝任何形式的转载,如有违者,将保留追究权利!
用sysbench做测试,--max-requet=100000 --num-thread=8 --mysql-table-engine=innodb
dell 2950(8g ram) vs hp 360g5(8g ram),/home 均做成4块盘的raid1+0。
从下面的测试结果来看,hp要比dell的机器稍好一点,但如果只是用dd做简单文件拷贝测试,则dell会比hp的机器快。
Submitted by yejr on 周五, 2008/07/18 - 10:22
前言:随着数据量的增加,备份问题会越来越突出,尤其是当前没有免费热备工具的情形下。本文针对 InnoDB 的备份需求,采用商业的 ibbackup 以及免费的 mysqldump 2种工具进行对比,让大家有个大致的了解。
1. 准备
ibbackup 是 InnoDB 提供的收费工具,它支持在线热备 InnoDB 数据,主要有以下特性:
- * Online backup of InnoDB tables — the backup takes place entirely online, without preventing queries or updates.
- * Online backup of MyISAM tables — during the backup of InnoDB tables, read and write access is permitted to MyISAM tables. While the MyISAM tables are being copied, updates (but not reads) to the MyISAM tables are precluded.
- * Compressed backups — the backup of InnoDB files can be compressed at various levels, saving as much as 70% or more of the storage required.
- * Partial backups — you can selectively backup all or only some of your InnoDB tables.
- * High performance — backup time is comparable to file copy, applying logs for recovery is even faster.
- * Unlimited database size — no practical limit to database size or number of tables.
- * Broad platform support — runs on Linux, Windows and leading Unix platforms.
Submitted by yejr on 周四, 2008/07/17 - 12:18
引言:在以前,我总是习惯用 INT UNSIGNED 来存储一个转换成Unix时间戳的时间值,认为这样做从索引,比较等角度来讲,都会比较高效。现在我们来对比下 TIMESTAMP 和 INT UNSIGNED 以及 DATETIME 这3种类型到底谁更好。
1. 准备
创建一个测试表:
mysql> CREATE TABLE `t` (
`d1` int(10) unsigned NOT NULL default '0',
`d2` timestamp NOT NULL default CURRENT_TIMESTAMP,
`d3` datetime NOT NULL,
KEY `d2` (`d2`),
KEY `d1` (`d1`),
KEY `d3` (`d3`)
);
然后创建一个存储过程填充数据:
Submitted by yejr on 周二, 2008/07/15 - 09:37
Submitted by yejr on 周一, 2008/07/14 - 14:23
假设有个myisam表:tbl,为了备份方便,直接把 frm 和 MYD 文件拷贝到其他目录。在还原时,就需要重新下创建索引,只需要执行以下命令:
mysql> REPAIR TABLE `tbl` USE_FRM;
即可根据 frm 和 MYD 文件,产生一个新的 MYI 索引文件。这在索引文件较大时备份还原比较有用。
另外,mysql 5.1起开始支持 分区(partition),但是如果表的某个子分区索引坏了,用 REPAIR TABLE 却无法修复,咋办呢?
其实很简单,利用 myisamchk 工具就可以修复了,例如:
path/myisamchk -rq tbl#P#P2
- check record delete-chain
页面
最近评论