Re: read performance much slower than write

[email protected] (Peter T. Breuer)
Newsgroups gmane.linux.enbd.general
Message-ID <[email protected]>
Hoffmann Christian <[email protected]> wrote:
> I try to format the mesurement new and split it to Output and Input:

OK, thanks! But what happened to the byte-by-byte measurements?

> Version  1.03
> ------Sequential Output------ 
> -Per Chr- --Block-- -Rewrite- 
> Size K/sec %CP K/sec %CP K/sec %CP 
> 4G             72113  42 18330   9           

There's still a "per chr" there that I don't understand! Originally I
thought the 72MB/s was under that heading, so I'm a little suspicious!
What was the "per chr" measurement for write?

> --Sequential Input- --Random-
> -Per Chr- --Block-- --Seeks--
> K/sec %CP K/sec %CP  /sec %CP
>           11445   3 755.9   2

Again, I'd like to know the "per chr" value, for reassurance.


> If I read this right it says that Block write is 72MB/s, rewrite is 18MB/s
> and read Block is 11MB/s.

Possibly.

> I cannot find out why the read performance is so slow compared to the write
> performance.

But you haven't looked!

What are the measurements on the enbd device itself?  On raid alone?  On
raid over enbd? On reiserfs over enbd. We have the measurement on
reiserfs over raid over enbd.

> I also tried the test on the servers local harddisk and it is about 90MB/s
> write and 140MB/s read.

OK, now to find the reason, one must measure the contribution made by
each component.

Peter
lmpx.com only provides a reader for public news (NNTP) servers. It is not affiliated with the servers or forums shown here and is not responsible for the content of articles, which is written by their respective authors.