Visualization with stat data

[raffis]

I want to create graphs with stat data, but I've realized that this is impossible as far as I understood kibana correctly.
@see similar topics: How to get latest entry for all users

I just wrote a big nagios performance data filter for logstash, and just realized I can't create visualizations which make sense.
For example:
How can I create a metric number with the current RAM Usage over multiple host?
How can I create a line chart with RAM Usage over one day (x=Date Histogram, y=Ram Usage)?

This seems impossible because I always get the sum of the stat data which clearly doesn't make any sense.

Kibana Filter: host:vmhost* AND nagios.service.raw:"VMware ESXi Host Memory"

Data example:
{host:vmhost1.example.org, nagios.service:"VMware ESXi Host Memory": 133,000, nagios.unit:"MB"}
{host:vmhost2.example.org, nagios.service:"VMware ESXi Host Memory": 70,000, nagios.unit:"MB"}
{host:vmhost1.example.org, nagios.service:"VMware ESXi Host Memory": 134,430, nagios.unit:"MB"}
{host:vmhost2.example.org, nagios.service:"VMware ESXi Host Memory": 69,545, nagios.unit:"MB"}
{host:vmhost1.example.org, nagios.service:"VMware ESXi Host Memory": 152,430, nagios.unit:"MB"}
{host:vmhost2.example.org, nagios.service:"VMware ESXi Host Memory": 65,444, nagios.unit:"MB"}

[tbragin (Tanya Bragin)]

So, it's true that you can't get the current (latest) value - there is no aggregation exposed in Kibana that will give you that. Feel free to +1 this ticket: https://github.com/elastic/kibana/issues/678

However, you should be able to get the overtime chart you described by using the "Avg" aggregation on the Y-axis. In fact, that is exactly what we do in Topbeat. Note that in this case, the indexed value was in bytes, and I used a field formatter to display it in MB.

[raffis]

@tbragin

However, you should be able to get the overtime chart you described by using the "Avg" aggregation on the Y-axis. In fact, that is exactly what we do in Topbeat. Note that in this case, the indexed value was in bytes, and I used a field formatter to display it in MB.

Yeah this is working fine with one server but not with multiple server.
With multiple server I get the average between them which is actually wrong.
What I need is the average of each and then the sum() of that.