The Most Common Step Is Often the Wrong Step

Figure 1. Basic Connectivity Diagram

The first issue with subscriber routers is that they are designed to be really cheap; the lowest bidder often wins. 

Second, the subscriber routers are extremely complicated because of their many roles:

1. Communicate with the primary network
2. Control Upstream power
3. Perform Upstream Muxing
4. Demux the Downstream traffic
5. Assign IP addresses to all devices
6. Manage input and output traffic to all devices
7. Perform Wi-Fi hub duties
8. Manage subscriber settings
9. Store performance data
10. Store diagnostic data

Cheap and complicated are rarely recipes for success.

InvisiLight® Solution for Deploying Fiber

Go to Market Faster. Speed up Network Deployment

Episode 10: Fiber Optic Closure Specs Explained…

Food for Thought from Our 2022 ICT Visionaries

This is why a power cycle often helps, at least short term. The routers are processing so many things at the same time that the firmware gets out of sorts or some buffer becomes full. A power cycle resets the firmware and empties all the buffers. With the return to "sanity", the service works well again.

The problem is that whatever overwhelmed the router could be a rare combination of freak events, or something common enough to overwhelm the modem again soon after the power cycle. An example I have seen is one device that keeps asking for a new IP address even though it has one. A television at one subscriber asked for a new IP address until the router reached its preset limit. Then the router would punt and stop delivering traffic to anything. The power cycle worked for a few hours until the limit was reached again.

Yeah, again: so why is power cycling not a good idea? 

Sure, it often works. The reason is that when you power cycle, the performance data and diagnostic data that exists only in the subscriber router is erased. You can’t tell what factor(s) overwhelmed the router. 

Why is this a big deal? Because the root cause could be anything from a shorted cable pair to bit storms. Most routers hold enough data to isolate to the appropriate layer and find the root cause. (See Figure 2.)

Figure 2. Example of router data that resides only in the router.

Problem-Solving Options

Am I advocating the end of power cycling? Of course not. Getting the service up as quickly as possible and moving on to the next call is the mantra for customer service centers. 

Instead, this is what I am advocating:

Where remote router access is supported

• Access the router.
• Check a few statistics.
• Look for "insanity" within the stats.
• If "sane", perform the power cycle; if not "sane", escalate to Tier 2 for deep analysis.

I get that "sane" is not specific enough to be helpful, but the required information will not fit into a magazine column. Using the example above, "insane" would be a residence where 999 IP addresses have been assigned.

Where remote router access is not supported

• Query the subscriber as to what the LEDs are doing
• Look for obvious LED conclusions
• Power cycle the router

What are obvious conclusions? For example, if the sync LED is flashing slowly, there is no continuity between the router and the primary network. There could still be Layers 2-7 issues, but Layer 1 must be fixed first. 

When the trouble is a repeated report
Ask the subscriber how long the service worked after the last power cycle.

• If less than one day, escalate to Tier 2 without power cycling.
• If more than 1 day, dispatch looking for physical or noise issues.
• If a power cycle does restore service, do NOT assume that anything has been fixed.

To wrap up, let me share a personal story. The DNS server for my office Internet would point only to Google IP addresses. I erased the history for my browsers, pointed to alternate DNS servers, power cycled the computer, flushed the DNS settings — and nothing worked. 

I called into the cable company, and what worked? Power cycling the router. It has been up for weeks.