| 2016-09-21 10:17:10 |
Patrick Petit |
description |
Into:
http://plugins.mirantis.com/docs/i/n/influxdb_grafana/influxdb_grafana-0.10-0.10.0-1.pdf
we have recommendation:
An average setup would
require a quad-core server with 8 GB of RAM and access to a 500-1000 IOPS disk.
As we on large environment >150 nodes disk subsystem is overloaded.
The utilization rate is about 98-99% which does not allow the system to work with data.
Current IOPS load is near 450 IOPS.
Additional we discovered that some historical data is not correct or missed for some periods as well due to disk subsystem was not able to save(receive) data from cluster nodes.
I'd like to recommend to change "500-1000 IOPS disk" to enterprise SSD disks. |
Into:
http://plugins.mirantis.com/docs/i/n/influxdb_grafana/influxdb_grafana-0.10-0.10.0-1.pdf
we have recommendation:
An average setup would
require a quad-core server with 8 GB of RAM and access to a 500-1000 IOPS disk.
As we on large environment >150 nodes disk subsystem is overloaded.
The utilization rate is about 98-99% which does not allow the system to work with data.
Current IOPS load is near 450 IOPS.
Additional we discovered that some historical data is not correct or missed for some periods as well due to disk subsystem was not able to save(receive) data from cluster nodes.
I'd like to recommend to change "500-1000 IOPS disk" to enterprise SSD disks.
DocImpact |
|
