Customers are constantly pinging our edge router, ns, or mail server. Recently we had a flood of "your network is down" calls because customers where pinging a well known site we host. That site decided to block all ICMP (don't start with me). We have since tweeked there filters to allow certain ICMP things thru. End users (read those that write the checks) seem to put a lot of stock into the ping and trace route values. Certainly the DSL customers do. Yet, I am seeing the "net" becoming more asymetric, and those ICMP ECHO / ECHO-REPLY packets aren't taking nearly the same path as they used to. jmbrown On Mon, Dec 18, 2000 at 12:12:15AM -0800, Bill Woodcock wrote:
On 17 Dec 2000, Sean Donelan wrote: > Most network providers ping their routers for network status. Several > providers even track RTT to detect changes. But very few customers > connect to routers.
I'd disagree with this, at least from what I've observed of our customers... I see a fairly steady stream of ICMP directed to the loopback and tail-circuit interfaces of our core and tail-circuit routers from customer address space, and if I were to guess, I'd say that 10%-15% of our customers were using some sort of ICMP-based uptime-monitoring packages which are looking at their Internet connection, among other things. From talking with them, I think most of them are using them to monitor WAN and VPN link uptime, and that they just throw our router into the list as an afterthought.
-Bill