Mention that more reads happen during backup and DR.

This commit is contained in:
A.J. Beamon 2018-03-19 12:41:51 -07:00
parent 423ffb7acb
commit 6513287630
1 changed files with 1 additions and 1 deletions

View File

@ -104,5 +104,5 @@ A lot of things affect the simple first-order model of performance you see here.
* Not all CPUs are the same speed.
* To keep up with the performance modeled above, your disk subsystem will need to do a little over 1 IOPS per write, and about 1 IOPS per (uncached) read.
* Network performance tuning at the operating system level can be very important for both latency and throughput, especially in larger clusters.
* Running DR and/or backup requires applying each mutation multiple times and will reduce throughput.
* Running DR and/or backup requires applying each mutation multiple times and then reading those mutations from the database. Using either feature will reduce throughput.
* See our :ref:`known-limitations <known limitations>` for other considerations which may affect performance.