Graphite module performance issues

I am going to use graphite to replace MRTG.
I setup graphite (graphite server) and icingaweb2 with graphite module (icingaweb2 server) in 2 different servers.
check-nwc-health is used to pull the interface usage from the target devices
This is the command I use
‘/usr/lib64/nagios/plugins/check_nwc_health’ ‘–hostname’ ' ‘–community’ ‘public’ ‘–mode’ ‘interface-usage’ --name ‘(Ethernet)|^(Port-channel)’ --regexp ‘–selectedperfdata’ ‘traffic’ ‘–report’ ‘short’

It will check all interfaces and port-channels of the device.
There are about 200 interfaces within this check.

When the graphs are loaded within icingaweb2 by graphite module,
There are multiple httpd (apache) processes running with cpu load 15%+ each on the icingaweb2 server.
The graphite server also shows similar httpd process cpu usage.

Is that mean 1 graph mean 1 httpd process?
If I have 200 interfaces graph in one page, will there be 200 httpd process for each reload?
At the moment, there is only 1 users (me) accessing the pages.
Therefore, the servers can still handle the load.
But there will be multiple users access the page in the future which I dont think the server can sustain the load.

does those graph cache by graphite-cache somewhere which can be reduce the load? or
Do I need to separate the interface usage into multiple check to prevent too much graph load within the same page?