This topic has been archived. It cannot be replied.
-
工作学习 / 学科技术讨论 / 一个 9000GB 的SQLServer OLTP数据库,有什么好的备份方法。俺采用一周一次FULL备份,备份文件被平均分割成30个230GB compressed 文件,分布在不同drive上。每天做differetial BACKUP和日志 backup。有无更好的办法?数据库每天archive,但它的基本数据就这么大9000 GB。
-houseless(糊涂);
2009-11-22
(#5698028@0)
-
俺的问题是,那个FULL备份要TAKE 7 小时,这还不算拷贝到磁带的时间。
-houseless(糊涂);
2009-11-22
(#5698048@0)
-
What is your objective? What is your backup strategy?
-hkchan(0);
2009-11-22
(#5698168@0)
-
how can you recover it? what's your rto? copy another set of disks. and then run backup from there.
-saper((㊣金旺旺㊣));
2009-12-5
(#5728653@0)
-
Did you just backup your database once a week?
-hkchan(0);
2009-11-22
(#5698185@0)
-
full backup once a week, incremental backup everyday. typical strategy.
-coolmao(酷猫*Arrakis);
2009-11-22
(#5698192@0)
-
My bad! I missed the differential backup part.
-hkchan(0);
2009-11-22
(#5698195@0)
-
If money is not an concern, then try split-mirror backup. Backup your full database to another storage set.
-hkchan(0);
2009-11-22
(#5698211@0)
-
Depends on your RPO and RTO, as hkchan said, if you don't care the cost, disk array based backup is the best plus the traditional backup software TSM/Netbackup etc.
-ddlldd2000(hunting);
2009-11-22
(#5698796@0)
-
谢谢,俺已经split backup to30 files,道是在考虑增加一个SAN CONTROLLER专门给备份。MY objective就是想把备份时间加快到4--5小时。俺准备split backup from 30 files to 50 files试试,再者看看有无好的第三方软件。
-houseless(糊涂);
2009-11-22
(#5698928@0)
-
But with split mirror, it only takes a few minutes to offline your backup disks array. Again, if you have a lot of money, you can have a few backup disks arrays and take whatever time you need to backup your data to tape.This was how we backup our financial systems but I can't recall its size.
-hkchan(0);
2009-11-22
{74}
(#5698985@0)
-
牛大的数据库,这样备份迟早会累死的,不能archive出一部分吗?
-sunday8(每个年头都有新的梦景);
2009-11-24
(#5701643@0)
-
Get a 3rd party backup tool900+GB is small in my production, we have 7 dbs which are 1.5TB in average. We use RedGate sqlbackup tool to do the backup (weekly full backup + daily diff backup + 20 min t-log backup)
For full backup, it usually takes 4 to 5 hrs.
Hope this helps !
-yh_abc(yh_abc);
2009-11-30
{256}
(#5716443@0)
-
Wrong reply
-yh_abc(yh_abc);
2009-11-30
{651}
(#5716462@0)
-
1. logshipping 在standby server 端备份。 2. partition table在file group的不同文件上, 只备份当前update的partition. 3. sql 2008 compression 备份速度快因为write block减少。 4. split mirror硬件实现。5. 第三方工具备份效率高。
-walkthrough();
2009-12-1
(#5719660@0)
-
谢谢,1.公司不打算买机器for logshipping 2. 全数据库就三个巨大partition tables, 在三个filegroup里面,可以考虑filegroup在分多一些 4.俺需要问storage team. 5正考虑lite-speed,3。使用了compression 不行。
-houseless(川岛房子);
2009-12-10
(#5737842@0)
-
说个高端平台的技术:快速拷贝。IBM DS8000磁盘系统里面有个技术叫FlashCopy, 即在瞬间完成所有磁盘的拷贝,你的9TB也许不到10分钟即可拷贝完毕。IBM的DB2联机支持这种快速拷贝模式,当然将数据库停掉10分钟很适合做Full Backup.
当然这种高端磁盘还包含很多其他的技术,例如硬件级别的远程实时备份等。唯一的问题就是采购成本比较高。
-ready4u(就等你了);
2009-12-5
{178}
(#5728544@0)
-
flash copy can be completed in seconds. but it has performance impact on the production DB.
-saper((㊣金旺旺㊣));
2009-12-5
(#5728665@0)
-
ANY backup has performance impact to PRD DB, disk array based PIT (Point In Time) copy has the smallest impact.
-ddlldd2000(hunting);
2009-12-6
(#5730436@0)
-
一般FlashCopy之后,硬件会在后台拷贝所有数据。但它有个选项,即只拷贝被修改的原始数据。如果数据库不是修改频繁,性能影响应该不大。
-ready4u(就等你了);
2009-12-7
(#5730912@0)
-
you don't have to stop database to do the full backup with Flashcopy.
-ddlldd2000(hunting);
2009-12-6
(#5730428@0)
-
谢了,可以考虑,每天停SQLServer十分钟,用此技术备份全部的数据文件。
-houseless(川岛房子);
2009-12-10
(#5737840@0)
-
ds8000 还算高端的? 快照技术备份时会很影响生产卷,使用EMC BCV之类的才行,不过楼主单位估计不肯出钱买高端的设备。
-vmale2(游玩人生);
2009-12-11
(#5741591@0)
-
How long to restore?
-sm08(南泥湾);
2009-12-24
(#5762942@0)