User:Caleb Cooper/sandbox: Difference between revisions
Content deleted Content added
Caleb Cooper (talk | contribs) No edit summary |
Caleb Cooper (talk | contribs) Replaced content with "# one #:<pre class=scroll> #::some stuff #::some more stuff</pre> # two" Tag: Replaced |
||
| Line 1: | Line 1: | ||
# one |
|||
<pre> |
#:<pre class=scroll> |
||
#::some stuff |
|||
We have that issue on Unity because Data Reduction (Compression/Deduplication) works in 8 KB blocks. when we have volume with large numbers of writes smaller than 8 KB have to backfilled from the volume (a read to the backend), to get enough data to compress, which causes extra load on the CPU. The average write size masks the number of small writes to a certain extent, because any large writes can |
|||
</pre> |
#::some more stuff</pre> |
||
# two |
|||
<pre class="scroll"> |
|||
We have that issue on Unity because Data Reduction (Compression/Deduplication) works in 8 KB blocks. when we have volume with large numbers of writes smaller than 8 KB have to backfilled from the volume (a read to the backend), to get enough data to compress, which causes extra load on the CPU. The average write size masks the number of small writes to a certain extent, because any large writes can |
|||
</pre> |
|||
<pre class="inline"> |
|||
this |
|||
that |
|||
the other |
|||
</pre> |
|||
<pre class="blue-white"> |
|||
We have that issue on Unity because Data Reduction (Compression/Deduplication) works in 8 KB blocks. when we have volume with large numbers of writes smaller than 8 KB have to backfilled from the volume (a read to the backend), to get enough data to compress, which causes extra load on the CPU. The average write size masks the number of small writes to a certain extent, because any large writes can |
|||
</pre> |
|||
<pre class="blue-white scroll"> |
|||
We have that issue on Unity because Data Reduction (Compression/Deduplication) works in 8 KB blocks. when we have volume with large numbers of writes smaller than 8 KB have to backfilled from the volume (a read to the backend), to get enough data to compress, which causes extra load on the CPU. The average write size masks the number of small writes to a certain extent, because any large writes can |
|||
</pre> |
|||
<pre class="blue-white inline"> |
|||
this |
|||
that |
|||
the other |
|||
</pre> |
|||
<syntaxhighlight lang=text highlight="3" class="blue-white"> |
|||
#!/bin/bash |
|||
fileCount=800000 |
|||
b=0 |
|||
while [ $b -le $fileCount ] |
|||
do |
|||
echo B = $b |
|||
dd if=/dev/urandom of=/mnt/nfs/file$b.txt bs=2k count=1 |
|||
((b++)) |
|||
done |
|||
</syntaxhighlight> |
|||
Revision as of 19:29, 15 July 2020
- one
- some stuff
- some more stuff
- two