z/TPF TCP/IP Keepalives flooding the network

Today on a system we have 125,000 socket blocks defined, and we are reaching a peak of 90,000 active sockets through the day. We run with Keepalive active so we can ensure they are still valid, and currently have the TCPALIVE interval set to 360 seconds. While we won't have to send the keepalive for all 90,000 active sockets, there will be times where a lot of keepalives will be sent, maybe 20,000 or more in a given interval. The keepalive processing in CTLV does give up control after every 500th socket checked by doing a DLAYC, but going through the entire table is going to happen very fast. We have seen instances where keepalive processing seems to have exacerbated potential network issues, by flooding the network with a significant number of packets in a very short time frame.

We're wondering if there should be a limit on the number of keepalive messages sent at one time. An easy solution may be to change the DLAYC in CTLV to an EVNTC/EVNWC with a delay of a second. This may not be the best solution as it would delay every 500 sockets, regardless of if the keepalive is sent or not, but may not be an issue. A more elegant solution may be to add an additional counter, and only do the second delay if a specified number of keepalives (perhaps another CTK2 value) have been sent. Either way would limit the number of keepalives sent within a given time frame to a manageable and known amount.