While using the custom optimizer policy defined for Virtual Circuit (VC), the Interface graphs are showing different utilization bandwidth, than Real-Time monitoring.
Real-time value reaches approximately 200Mbps.
Interface throughput graphs are showing around 800Mbps when the VC is enabled.
The External Monitoring using the Cacti tool is showing similar values of 800-900 Mbps.
In Exinda logs, you may notice the following entry:
[statsd.NOTICE]: alarm 'exinda_cpu_indiv': alarm for event exinda_cpu_indiv rising clearing, but no event sent
This article provides details about possible reasons for such a difference in bandwidth.
In this particular scenario, the VC policy was created to handle the backend operation of AWS DataSync service.
There is an agent running in the local datacenter behind Exinda that reaches out file share servers. The DataSync agent then uploads data to AWS over HTTPS.
Because of the AWS and agent tool design, the connections are being opened concurrently, reaching the Packets Per Second (PPS) limit.
The Exinda device in place (4062) cannot handle such a number of packets (over 290.000 per second) due to the device's bottleneck and hardware specs. Such hardware and software limitations are causing Bandwidth reporting discrepancy.
The Exinda 4062 is only capable of processing 200.000 PPS.
It's recommended to purchase a hardware device that is capable of processing bigger PPS values, such as Exinda 8063.
Please contact GFI Sales for available hardware options.