Network routing

Off Topic
1

We’re still having issues with Tax Connect connectivity. When I run the Test Connection in Company Config, it fails about 20% of the time.

Some crude diagnostics I’ve tried include:

  • Pinging the Avalara endpoint (avatax.avalara.net)

    • Returns the IP, but no ping response. (This is not unexpected)
    • The IP varies from run to run, and appears to be on Amazon Web Services

  • TCPView (A crude viewer of active TCP connections) on App Server.

    • When the Test connection fails, a connection between the App Server and Avalara is shown, but with the connection status of SYN_SENT. This indicates the remote computer(Avalara) never responded.

  • Tracert from App Server to avatax.avalara.net

    • Tracert ends (after taking the full default 30 hops ) prior to reaching avalara
    • Increasing the max hop count to 100, and the trace ends prior to reaching avalara
    • The first two hops in the trace are IP addresses owned by us. But after that, they all appear to be other companies (sgns.net, quest.net, centurylink.net, etc …)

  • Tracert from my home PC (via my home ISP) to avatax.avalara.net

    • Tracert reaches avalara in just two hops

My questions:

  1. Is it odd that the IP address of a URL is constantly changing? I understand that there have to be many IP’s to handle the traffic, but shouldn’t there be some thing like DHCP that remembers the address for a client? And does my local system remember the IP in its DNS cache, such that successive uses of the same URL would want to use the same IP add?

  2. Am I barking up the wrong tree when I think the tracert should only be a couple of hops once It’s past my fire wall (like I see when tracing from my home ISP)

  3. Is there a way to improve the connectivity to a remote host whose IP is constantly changing?

2

Routing by your ISP can be a factor here.

  1. URL to IP mapping is done by DNS. There are settings such as round robin, load balancing which means that the answer can be different on each query. Factor such as the TTL (time to live) can control whether caching holds onto that answer for mins, hours, or not at all.

  2. Tracert is a good indicator - but like you found if there are hops on the path that are set to not respond to pings it gives you those blanks. Doesn’t mean that real traffic isn’t passed though, just because it didn’t respond to the ping.

  3. Yes - but not something you can do. On a home ISP, asking for these kinds of changes would be fruitless. The front line tech you get through to would likely not understand, or have a decent escalation path to get it resolved. On a proper business ISP though, this is definitely a conversion that you can have. Show them the tracert from home, and work and explain the issues that you’re facing. They will likely have multiple redundant links across their network and out to the public internet. They should have the ability to control your traffic onto a more optimal route if you are having problems.

2 Likes