Oracle’s redo log is a frequent source of intensive IO and, by design, Oracle sessions will normally wait for
redo log IO to complete. However, the nature of redo log IO operations is essentially optimal for magnetic
disk but least suitable for solid state disk. Redo log IO consists almost entirely of sequential write operations,
which are the worst case for solid state disk and the best case for magnetic spinning disk. Placing
redo logs on solid state disk is generally not recommended
When IO bottlenecks are limited to specific segments (tables, partitions, or indexes) then relocating those
segments to solid state disk will almost always be effective in optimizing IO to those segments, regardless of
the type of IO to which the segments are subjected
from guy harrison blog : http://guyharrison.squarespace.com/blog/2011/10/27/using-flash-disk-for-redo-on-exadata.html
New Tool: xstack - Completely Passive eBPF Linux Stack Profiling Without
Any Tracepoints
-
I just added another tool into my 0x.tools toolset: xstack is a
lightweight, completely passive stack profiler for Linux that uses eBPF
task iterators to s...
Acum 3 zile
Niciun comentariu:
Trimiteți un comentariu