I think the API should be improved to allow for a Connection Timeout and a Scan Timeout. Currently a single timeout is used for both (v 7.8.0.141).
If scanning large files it can take up to 20 or 30 seconds for the scan to complete. If connecting to a host, the connection should be very quick 1-5 seconds. It seems wrong to use the same timeout for both of these operations, especially when considering the load balancing api.
Here is a rough overview of how the load balancing API feature works to illustrate why this is important.
The SPE load balancing api allows the user to pass in a list of hostnames and it will attempt to connect to one of the hosts randomly until it gets back a message saying the host is busy, or the read timeout is reach (see above). It will then attempt to connect to another server in the list of hosts. After five retries, the API will return an exception for MAX_TRIES_REACHED.
Imagine how long this will take if the timeout is 30 seconds(needed for large files) and one or more hosts is offline for maintenance! This makes the API unusable for my use case so we will be switing to using a Load Balancer VIP.