关于ASSM,给你一段oracle公司最权威的专家的原话:Auto segment management just automates the process by which you used to have to manually tune using multiple freelists. In the past, you would have had to of determine the number of concurrent modifiers of data (inserts/updates that happen exactly at about the same time) and configure multiple freelists in order to satisfy these concurrent requests in a timely fashion.With auto management you don't have to do this anymore. It'll not distribute across multiple blocks any more then multiple freelists would have in the past -- only you need not waste your time and energy attempting to determine the exact correct number of freelists to have.This new feature is more efficient then multiple freelists:<quote> Free lists have been the traditional method of managing free space within segments. Bitmaps, however, provide a simpler and more efficient way of managing segment space. They provide better space utilization and completely eliminate any need to specify and tune the PCTUSED, FREELISTS, and FREELISTS GROUPS attributes for segments created in the tablespace. If such attributes should be specified, they are ignored. </quote> a very informal, quick and dirty test I've done shows that automatic segment space mgmt gives us almost all of the benfit of finely tuned freelists -- without the pain of finely tuning freelists (something that is virtually impossible to do in the "real world" anyway).Go for it -- on a heavy OLTP type table -- this is very beneficial.
感谢楼上诸位兄弟,小结下,只说性能,管理上优势显而易见: 1. Bigfile在某些可能影响性能的方面做的比Smallfile Performance of database opens, checkpoints, and DBWR processes should improve if data is stored in bigfile tablespaces instead of traditional tablespaces. However, increasing the datafile size might increase time to restore a corrupted file or create a new datafile. -- http://download.oracle.com/docs/cd/B19306_01/server.102/b14220/physical.htm#i14068 2. ASM可以改善性能,由上面天黑的回复及 ASM:A vertical integration of both the file system and the volume manager built specifically for Oracle Database files. It extends the concept of stripe and mirror everything to optimize performance, while removing the need for manual I/O tuning. -- http://download.oracle.com/docs/cd/B28359_01/server.111/b28318/glossary.htm#CNCPT1995 3. 不过上面都是ORACLE自己说的,有时候也当不得真。 4. 所以有没有人明白里面的实现机理或给出测试数据。 5. 此帖200分,哪边给建议都给分,明天结帐,多谢!
smallfile在管理方面更有靈活性。
如果一個tablespace要儲存巨大的數据(TB級別的),那麽就用bigfile吧,因爲用smallfile來保存的話,要用很多個smallfile,這樣會增加checkpoint時的IO操作。ps.對ASM還不是很熟悉,不討論了。
我想知道的是:
1. ASM对性能有没有什么影响,管理成本是比以前多了还是少了。
2. BIGFILE和SMALLFILE楼上也有兄弟说了,SMAILLFILE的CHECKPOINT有增多,可能比BIGFILE开销大。
3. 网上还有贴说创建BigFile,就会有很大的inode locking issues,使用单个TEMPFILE会影响查询性能等。俄想要个相对完整的输出
tune using multiple freelists. In the past, you would have had to of determine the
number of concurrent modifiers of data (inserts/updates that happen exactly at about the
same time) and configure multiple freelists in order to satisfy these concurrent requests
in a timely fashion.With auto management you don't have to do this anymore. It'll not distribute across
multiple blocks any more then multiple freelists would have in the past -- only you need
not waste your time and energy attempting to determine the exact correct number of
freelists to have.This new feature is more efficient then multiple freelists:<quote>
Free lists have been the traditional method of managing free space within segments.
Bitmaps, however, provide a simpler and more efficient way of managing segment space.
They provide better space utilization and completely eliminate any need to specify and
tune the PCTUSED, FREELISTS, and FREELISTS GROUPS attributes
for segments created in the tablespace. If such attributes should be specified, they are
ignored.
</quote>
a very informal, quick and dirty test I've done shows that automatic segment space mgmt
gives us almost all of the benfit of finely tuned freelists -- without the pain of finely
tuning freelists (something that is virtually impossible to do in the "real world"
anyway).Go for it -- on a heavy OLTP type table -- this is very beneficial.
这里没有推荐的意义吧,突出的bigfile的好处
也就是在storage上,不需要多个datafile叻,但是我决定这个对于文件管理来说是好处,毕竟只对应一个文件,当然比多个文件好管理,不过性能上倒是比不到smallfile哟,毕竟small file又方式分散到IO如果处理的好的话。而且对于管理来说如果用ASM的话。我们也不需要对文件直接管理哟所以这里smallfile解决的storage capacity的问题,并不是解决better-performance的问题的,
不应该把smallfile/bigfile和ASM进行比较,比较方向不同,没有可比性。也可以见我在这个贴上的回帖
http://topic.csdn.net/u/20090729/09/9b958ebf-533e-4200-bc4e-3c1c11c4b329.html?85728
这个和bigfile就没有关系叻。
1. Bigfile在某些可能影响性能的方面做的比Smallfile
Performance of database opens, checkpoints, and DBWR processes should improve if data is stored in bigfile tablespaces instead of traditional tablespaces. However, increasing the datafile size might increase time to restore a corrupted file or create a new datafile.
-- http://download.oracle.com/docs/cd/B19306_01/server.102/b14220/physical.htm#i14068
2. ASM可以改善性能,由上面天黑的回复及
ASM:A vertical integration of both the file system and the volume manager built specifically for Oracle Database files. It extends the concept of stripe and mirror everything to optimize performance, while removing the need for manual I/O tuning.
-- http://download.oracle.com/docs/cd/B28359_01/server.111/b28318/glossary.htm#CNCPT1995
3. 不过上面都是ORACLE自己说的,有时候也当不得真。
4. 所以有没有人明白里面的实现机理或给出测试数据。
5. 此帖200分,哪边给建议都给分,明天结帐,多谢!
checkpoint会提高,就有些不明白了。难道是如果多个datafile的话,checkpoint会去同步后比较然后在确认,还是???
ASM还可以把磁盘读写均衡的分摊到他的diskgroup中.
ASM还有一个优点,在RAC环境下做归档或者快闪的时候,RAW环境下一般需要NFS 挂载文件系统来解决.
有了ASM,可以把归档空间还有快闪目的地都指向ASM.ASM有一个大的缺点,我想这应该是一个很多生产系统都不敢用它的原因,就是它是基于ASM INSTANCE的,
如果这个instance挂掉了,那么所有数据都完了.
ASM的Oracle的小孩,却是含着无比的关注和优良的基因而生的。让我们一起看着它能够快速的成长为肆意在存储解决方案的领域里领跑的人物。