From 1af4701b854d14134e5b9fdba6ba73f5326c361b Mon Sep 17 00:00:00 2001 From: Nikita Konovalov Date: Tue, 15 Mar 2016 15:30:48 +0300 Subject: [PATCH] Updated volumes section in docs Improved description for cinder volumes usage mechanism. Described the volumes backed by distributed storage case. Change-Id: Ie5a436daaca9e5ed009aa131cf6487be3057182a Closes-bug: #1371360 --- doc/source/userdoc/features.rst | 12 ++++++++++++ 1 file changed, 12 insertions(+) diff --git a/doc/source/userdoc/features.rst b/doc/source/userdoc/features.rst index 4decbe9029..1d3c37eadc 100644 --- a/doc/source/userdoc/features.rst +++ b/doc/source/userdoc/features.rst @@ -42,6 +42,18 @@ A user can set how many volumes will be attached to each instance in a node group and the size of each volume. All volumes are attached during cluster creation and scaling operations. +If volumes are used for the HDFS storage it's important to make sure that +the linear read-write operations as well as IOpS level are high enough to +handle the workload. Volumes placed on the same compute host provide a higher +level of performance. + +In some cases cinder volumes can be backed by a distributed storage like Ceph. +In this type of installation it's important to make sure that the network +latency and speed do not become a blocker for HDFS. Distributed storage +solutions usually provide their own replication mechanism. HDFS replication +should be disabled so that it does not generate redundant traffic across the +cloud. + Cluster scaling ---------------