BCP慢的问题
<p>原来的系统是 SCO5 sybase11,数据库的设备文件全在阵列上面,由于阵列硬盘和服务器硬盘掉电坏掉,把原来每天以二进制格式BCP出的数据恢复到现新服务器上,新服务器是HPML370G5 系统是REDHAT4UP464位,SYBASE是12.5.4,4G内存。本地四块SAS盘做的RAID5,把原来的备份的二进制文件FTP到新服务器上面,发现小文件恢复的速度还可以,但是大文件特别慢,BCP 单个900M的数据文件要近二十个小时,怎么能够改进一些,表的索引和触发器都以删掉。<br/>服务器和sybase 配置参数如下<br/></p><p>os<br/>kernel.shmmax = 4294967296<br/></p>
<p>sybase<br/><br/>sp_configure "max memory",1300000<br/>go<br/>sp_configure "allocate max shared mem",1 <br/>go <br/>sp_cacheconfig "default data cache","1120M"<br/>go <br/>sp_configure "procedure cache size",240000<br/>go <br/>sp_configure "allocate max shared mem",0 <br/>go <br/>sp_configure "number of open objects",5000 <br/>go <br/>sp_configure "number of pre-allocated extent",20 <br/>go <br/>sp_poolconfig "default data cache","80M","4K"<br/>go<br/>sp_poolconfig "default data cache","160M","8K"<br/>go<br/>sp_poolconfig "default data cache","240M","16K"<br/>go<br/>sp_configure "number of open indexes",4000 <br/>go<br/>sp_configure "number of devices",25<br/>go<br/>sp_configure "tcp no delay",1<br/>go<br/>sp_configure "number of user connections",250<br/>go<br/>sp_configure "number of locks",90000<br/>go</p> 是否enable了bulk copy?<br/> <p>那个是 enable 的</p>
页:
[1]