single session I/O bandwidth
From: Ls Cheng <exriscer_at_gmail.com>
Date: Tue, 12 Mar 2013 00:01:06 +0100
Message-ID: <CAJ2-Qb9t9X65gTo24dxdoAfp_Em=qcK3p5m7J9sJU51NppgtiQ_at_mail.gmail.com>
Hi
I have this doubt for quite long time and asked qite a few people but seems that so far no one knows the reasoning.
Date: Tue, 12 Mar 2013 00:01:06 +0100
Message-ID: <CAJ2-Qb9t9X65gTo24dxdoAfp_Em=qcK3p5m7J9sJU51NppgtiQ_at_mail.gmail.com>
Hi
I have this doubt for quite long time and asked qite a few people but seems that so far no one knows the reasoning.
If I run a simple FTS query against a 8GB table for example I can observe a I/O throughput of around 30 to 40MB, if I run 2 session each get 30 to 40MB bandwidth too (aggregate 80MB/sec) however the storage array has more than that, the maximum throughput I got from an XIV array recently is 250MB read and 250MB write running 8 parallels.
What is the mechanism which limits a single session cannot have more than 30, 40MB throughput? Some sort of SAN QOS?
TIA
-- http://www.freelists.org/webpage/oracle-lReceived on Tue Mar 12 2013 - 00:01:06 CET