All traffic traversing the partner gateway is impacted regardless of type of traffic and path (direct to internet, handoff, via datacenter, via edge).
This will be seen as Over Capacity Drops from the gateway monitoring page in the VCO, and users may report packet loss during these times.
Capacity drops indicate the gateway is unable to process the amount of traffic it's receiving. A large burst of traffic may cause drops briefly and often isn't a concern, but continuous drops throughout high usage periods or throughout the day can indicate one of a few different things:
- The gateway is working fine, but it's receiving more traffic than it can handle.
- The gateway is not deployed correctly, and is dropping traffic even though it doesn't appear to be oversubscribed.
- The gateway is experiencing some other issue and requires deeper analysis.
As a first step, ensure the ESXi or KVM host and gateway VM have been deployed according to the Hypervisor Minimum requirements. This step is important to ensure maximum performance based on the configurations that have been tested. In addition, Broadcom cannot support performance issues on hypervisors or gateways that aren't deployed correctly, as we cannot guarantee or determine what the expected performance will be since we have not tested with configurations outside of these requirements.
If the gateway was recently upgraded from 4.5 or older to 5.2 or newer, we also need to ensure the VM Hardware Compatibility is upgraded. Some changes were made to how the gateways use CPU cores for different functions, which doesn't work properly if the VM Hardware Compatibility is something older like ESXi 5.5.
To summarise, VCG should be provisioned with the following ESXi settings for getting optimal and guaranteed performance -
If the above have been vetted and/or remedied and the problem persists, please open a support request for further assistance.