More on Channel Timeouts

I wrote recently about the mechanics of implementing channel timeouts in a payment switch application.  Today, I explained to someone in an e-mail how the process works in practice (I’ve done a small bit of masking from my original comments):

The OLS application does proactive management of endpoint channel connections to protect against the possibility of “hung up” lines.  We look for the presence of an application “response” message and – if none is received within a designated timeout period – we force-disconnect the line and re-establish it.  [Some authorization endpoints – SVS, for example – undertake similar practices from their end.]  The timeout period associated with [XXX] is 20 minutes.  Each channel connection keeps track separately.   In periods of low volume overnight, you will see occasional channel disconnect/reconnect cycles on the OLS logs.  The [XXX] channels are especially prone to this behavior because there is not a Network Message (a.k.a., ‘0800’) in the [XXX] transaction set to keep the channel active in the absence of customer-initiated transaction traffic.  This Network Message feature is known by various names:  “Echo”; “Are You There?”; “RUOK”; and “Heartbeat” are commonly used terms we see in authorizer specifications.