Re: 答复: Horizontally scaling a database

  • From: "Sandeep Dubey" <dubey.sandeep@xxxxxxxxx>
  • To: tomdaytwo@xxxxxxxxx
  • Date: Fri, 16 Mar 2007 15:49:58 -0400

<Oh and I like the performance metric you put out. 2 seconds to return
a row from a billion row table. You think that is realistic? >

It may not be realistic or may be realistic. I get 600 ms response
time on 1.4 billion row partitioned table with global index. So by
that benchmark I was shooting for 2 sec or at the worst 4 sec. If this
is not achievable with in 4 sec, we will scrap the idea and go with
SAN and a beefy server.

I am already working on the idea as Thomas suggested  - creating view
of union of tables in all databases in the network. ( 4 for testing)



To query the entire data store we used views that did unions across database

Performance will be terrible, but it does what you asked for.

Go out and get a bunch of Pentium III machinces surplus with 40GB hard
drives.  Load Oracle and create the identical database on each machine (just
vary the SID).  Put them all on a network and have a master machine whose
database is nothing but views and database links.


Other related posts: