monitoring graphs are cut

Good morning.
I am presenting detail in the observium the monitoring graphs are cut, in the processor it sees that it increases up to 98% on occasions. We execute the recommendations of the portal to have better performance but this detail continues to be presented. we currently have 2462 devices with all their sensors. In general the platform has been operating correctly until a couple of weeks ago. In which these cuts are presented.

Statistics DB size 1.57GB 
RRD size 105GB 
Devices 2462 
Ports 24020 
IPv4 Addresses 9253 
IPv4 Networks 4710 
Processors 6370 
Memory pools 1808 
Storage Entries 1683 
Disk I/O Entries 59 
HR-MIB Entries 4665 
Entity-MIB Entries 5461 
Eventlog Entries 2746992 
Sensors 7864