Comcast @Home service Uptime Statistics--Data

New! September, 2000: For comparison purposes, I've now added a page showing the uptime statistics for plainsboro.com's own network

This page contains graphs of the data I have collected month by month for the Comcast ewndsr1 head end. Follow this link for an explanation of what data was collected and how it was collected.

Follow this link to the end of this page for an explanation of how the graphs were created and how to interpret them.

December, 2000 statistics (99.73% overall uptime) Grade: A+

Executive Summary

Total number of successful pings: 8904 (out of 8928)
Total number of unsuccessful pings: 24 (out of 8928)
Resulting uptime percentage: 99.73 %
Longest period of continuous uptime: 206 hours, 15 minutes
Longest period of continuous downtime: 35 minutes
Number of unique downtime periods of 15 minutes or more: 1
Number of unique downtime periods of 30 minutes or more: 0
Number of unique downtime periods of 1 hour or more: 0
Number of unique downtime periods of 4 hours or more: 0
Number of unique downtime periods of 1 day or more: 0

November, 2000 statistics (99.77% overall uptime) Grade: A+ (BEST MONTH TO DATE!!!!)

Executive Summary

Total number of successful pings: 8620 (out of 8640)
Total number of unsuccessful pings: 20 (out of 8640)
Resulting uptime percentage: 99.77 %
Longest period of continuous uptime: 138 hours, 45 minutes
Longest period of continuous downtime: 20 minutes
Number of unique downtime periods of 15 minutes or more: 1
Number of unique downtime periods of 30 minutes or more: 0
Number of unique downtime periods of 1 hour or more: 0
Number of unique downtime periods of 4 hours or more: 0
Number of unique downtime periods of 1 day or more: 0

October, 2000 statistics (98.86% overall uptime) Grade: A-

Executive Summary

Total number of successful pings: 8826 (out of 8928)
Total number of unsuccessful pings: 102 (out of 8928)
Resulting uptime percentage: 98.86 %
Longest period of continuous uptime: 79 hours, 40 minutes
Longest period of continuous downtime: 1 hours, 0 minutes
Number of unique downtime periods of 15 minutes or more: 7
Number of unique downtime periods of 30 minutes or more: 3
Number of unique downtime periods of 1 hour or more: 1
Number of unique downtime periods of 4 hours or more: 0
Number of unique downtime periods of 1 day or more: 0

September, 2000 statistics (99.02% overall uptime) Grade: A

There was a period from about 6:45 PM until about 8:55 PM on Saturday, September 30 where Comcast's internal network was having problems keeping its customers from accessing the internet. So my "ping" program was correctly reporting the default router up, but in reality, access beyond the default router was down (I therefore changed the log for this period of 2 hours and 10 minutes to reflect this actual downtime).

That one outage accounted for nearly 1/3 of all of the downtime for the month of September. Otherwise, great job for the month! Thanks, Comcast! You get an "A" this month.

Executive Summary

Total number of successful pings: 8555 (out of 8640)
Total number of unsuccessful pings: 85 (out of 8640)
Resulting uptime percentage: 99.02 %
Longest period of continuous uptime: 96 hours, 45 minutes
Longest period of continuous downtime: 2 hours, 10 minutes
Number of unique downtime periods of 15 minutes or more: 8
Number of unique downtime periods of 30 minutes or more: 3
Number of unique downtime periods of 1 hour or more: 1
Number of unique downtime periods of 4 hours or more: 0
Number of unique downtime periods of 1 day or more: 0

August, 2000 statistics (99.12% overall uptime) Grade: A

There was a severe thunderstorm in Mercer/Middlesex county the evening of Tuesday August 1, causing an @Home outage starting around midnight. The service was back up by 1:30, and my guess is this is because some under-paid network administrator came into work at that time in the morning to reboot equipment and fix other problems as necessary--thanks for doing this work at that late hour!

A very puzzling and worrisome pattern began developing in Comcast's network beginning the weekend of August 18. At least twice during that weekend, Comcast ewndsr1 users experienced lengthy periods of what I can only describe as "network sluggishness". While the method I use in these pages to track internet connectivity (pinging the default gateway machine) continued to show the defualt router as "up", this network sluggishness in fact made the internet unusable. Symptoms include:

Telnet to a remote unix host provides an immediate "connected" message, but the login prompt does not appear for another 5 minutes
Several minutes required to download simple pages from any remote web site
Failed or lengthy DNS lookups (even within the home.{com,net} domains)
The unix version of the "traceroute" command requiring from three to seven minutes to complete a 13-hop trace (more than half of which are inside Comcast's network).

I have no explanation of this network slowness, but I'm disturbed by the pattern.

That problem aside, Comcast did an excellent job keeping its @Home users' network up during the month of August, 2000. Despite the fact that one of their downtime periods occurred at an unfortunate time for me personally, their 99.12% overall uptime for the month deserves a grade of "A"!

Executive Summary

Total number of successful pings: 8849 (out of 8928)
Total number of unsuccessful pings: 79 (out of 8928)
Resulting uptime percentage: 99.12 %
Longest period of continuous uptime: 76 hours, 20 minutes
Longest period of continuous downtime: 1 hour, 35 minutes
Number of unique downtime periods of 15 minutes or more: 7
Number of unique downtime periods of 30 minutes or more: 1
Number of unique downtime periods of 1 hour or more: 1
Number of unique downtime periods of 4 hours or more: 0
Number of unique downtime periods of 1 day or more: 0

July, 2000 statistics (97.55% overall uptime) Grade: B-

I was all ready to give Comcast at least a B+ for the month of July, 2000. Through Saturday, July 29 they had a tremendous record of 98.5% uptime for the month. But at the last minute starting at about 8:30 PM Sunday evening July 30, some very scattered downtime occurred, including this puzzling pattern of ups and downs between 8:30 PM and 1:30 AM Sunday evening:


	24.2.82.1 OK at Sun Jul 30 20:35:00 EDT 2000
	Cannot ping default gateway 24.2.82.1 at Sun Jul 30 20:40:01 EDT 2000
	24.2.82.1 OK at Sun Jul 30 20:45:00 EDT 2000
	24.2.82.1 OK at Sun Jul 30 20:50:00 EDT 2000
	24.2.82.1 OK at Sun Jul 30 20:55:00 EDT 2000
	Cannot ping default gateway 24.2.82.1 at Sun Jul 30 21:00:00 EDT 2000
	24.2.82.1 OK at Sun Jul 30 21:05:00 EDT 2000
	24.2.82.1 OK at Sun Jul 30 21:10:00 EDT 2000
	24.2.82.1 OK at Sun Jul 30 21:15:00 EDT 2000
	24.2.82.1 OK at Sun Jul 30 21:20:00 EDT 2000
	24.2.82.1 OK at Sun Jul 30 21:25:00 EDT 2000
	24.2.82.1 OK at Sun Jul 30 21:30:00 EDT 2000
	Cannot ping default gateway 24.2.82.1 at Sun Jul 30 21:35:00 EDT 2000
	24.2.82.1 OK at Sun Jul 30 21:40:00 EDT 2000
	24.2.82.1 OK at Sun Jul 30 21:45:00 EDT 2000
	Cannot ping default gateway 24.2.82.1 at Sun Jul 30 21:50:00 EDT 2000
	Cannot ping default gateway 24.2.82.1 at Sun Jul 30 21:55:00 EDT 2000
	24.2.82.1 OK at Sun Jul 30 22:00:01 EDT 2000
	Cannot ping default gateway 24.2.82.1 at Sun Jul 30 22:05:00 EDT 2000
	24.2.82.1 OK at Sun Jul 30 22:10:01 EDT 2000
	24.2.82.1 OK at Sun Jul 30 22:15:01 EDT 2000
	24.2.82.1 OK at Sun Jul 30 22:20:00 EDT 2000
	24.2.82.1 OK at Sun Jul 30 22:25:00 EDT 2000
	24.2.82.1 OK at Sun Jul 30 22:30:00 EDT 2000
	24.2.82.1 OK at Sun Jul 30 22:35:00 EDT 2000
	Cannot ping default gateway 24.2.82.1 at Sun Jul 30 22:40:00 EDT 2000
	24.2.82.1 OK at Sun Jul 30 22:45:00 EDT 2000
	Cannot ping default gateway 24.2.82.1 at Sun Jul 30 22:50:01 EDT 2000
	24.2.82.1 OK at Sun Jul 30 22:55:00 EDT 2000
	Cannot ping default gateway 24.2.82.1 at Sun Jul 30 23:00:00 EDT 2000
	24.2.82.1 OK at Sun Jul 30 23:05:00 EDT 2000
	24.2.82.1 OK at Sun Jul 30 23:10:00 EDT 2000
	Cannot ping default gateway 24.2.82.1 at Sun Jul 30 23:15:00 EDT 2000
	24.2.82.1 OK at Sun Jul 30 23:20:00 EDT 2000
	Cannot ping default gateway 24.2.82.1 at Sun Jul 30 23:25:00 EDT 2000
	24.2.82.1 OK at Sun Jul 30 23:30:00 EDT 2000
	24.2.82.1 OK at Sun Jul 30 23:35:00 EDT 2000
	24.2.82.1 OK at Sun Jul 30 23:40:00 EDT 2000
	Cannot ping default gateway 24.2.82.1 at Sun Jul 30 23:45:00 EDT 2000
	Cannot ping default gateway 24.2.82.1 at Sun Jul 30 23:50:00 EDT 2000
	Cannot ping default gateway 24.2.82.1 at Sun Jul 30 23:55:01 EDT 2000
	24.2.82.1 OK at Mon Jul 31 00:00:00 EDT 2000
	24.2.82.1 OK at Mon Jul 31 00:05:00 EDT 2000
	24.2.82.1 OK at Mon Jul 31 00:10:01 EDT 2000
	24.2.82.1 OK at Mon Jul 31 00:15:00 EDT 2000
	Cannot ping default gateway 24.2.82.1 at Mon Jul 31 00:20:00 EDT 2000
	24.2.82.1 OK at Mon Jul 31 00:25:00 EDT 2000
	Cannot ping default gateway 24.2.82.1 at Mon Jul 31 00:30:01 EDT 2000
	24.2.82.1 OK at Mon Jul 31 00:35:00 EDT 2000
	Cannot ping default gateway 24.2.82.1 at Mon Jul 31 00:40:00 EDT 2000
	Cannot ping default gateway 24.2.82.1 at Mon Jul 31 00:45:01 EDT 2000
	Cannot ping default gateway 24.2.82.1 at Mon Jul 31 00:50:00 EDT 2000
	Cannot ping default gateway 24.2.82.1 at Mon Jul 31 00:55:00 EDT 2000
	24.2.82.1 OK at Mon Jul 31 01:00:00 EDT 2000
	Cannot ping default gateway 24.2.82.1 at Mon Jul 31 01:05:00 EDT 2000
	Cannot ping default gateway 24.2.82.1 at Mon Jul 31 01:10:00 EDT 2000
	24.2.82.1 OK at Mon Jul 31 01:15:00 EDT 2000
	Cannot ping default gateway 24.2.82.1 at Mon Jul 31 01:20:00 EDT 2000
	24.2.82.1 OK at Mon Jul 31 01:25:00 EDT 2000

Note, however, the essentially solid blue line of "2's" in the scatter graph below for July. This is an indication that there were no periods of extended downtime, which earns Comcast a B- this month.

Executive Summary

Total number of successful pings: 8709 (out of 8928)
Total number of unsuccessful pings: 219 (out of 8928)
Resulting uptime percentage: 97.55 %
Longest period of continuous uptime: 70 hours, 15 minutes
Longest period of continuous downtime: 2 hours, 30 minutes
Number of unique downtime periods of 15 minutes or more: 21
Number of unique downtime periods of 30 minutes or more: 9
Number of unique downtime periods of 1 hour or more: 5
Number of unique downtime periods of 4 hours or more: 0
Number of unique downtime periods of 1 day or more: 0

June, 2000 statistics (88.98% overall uptime) Grade: D

Executive Summary

Total number of successful pings: 5182 (out of 5824)
Total number of unsuccessful pings: 642 (out of 5824)
Resulting uptime percentage: 88.98 %
Longest period of continuous uptime: 50 hours, 30 minutes
Longest period of continuous downtime: 26 hours, 30 minutes
Number of unique downtime periods of 15 minutes or more: 10
Number of unique downtime periods of 30 minutes or more: 7
Number of unique downtime periods of 1 hour or more: 6
Number of unique downtime periods of 4 hours or more: 5
Number of unique downtime periods of 1 day or more: 1

I created the graphs on this page using Microsoft's Excel spreadsheet from their "Office 2000" product. I selected to put the graphs in "scatter graph" format because it seemed to be the format that displays the data for the clearest interpretation. Each of the graphs represents one month of data (or nearly one month). Each point on the graph indicates that either the Comcast ewndsr1 head end was "up" (indicated by a point at the "2" level) or that it was "down" (indicated by a point at the "1" level. Ideally, then, what you'd like to see is a solid line at the "2" level and no points at all at the "1" level; this would indicate 100% uptime. Each point at the "1" level indicates an unsuccessful ping of the head-end box. Short lines at the "1" level indicate extended downtime periods.

Each month's graph is preceded by an "executive summary" of the month's statistics. I will try to make a commitment to continue to gather data, and to keep the data current from month to month (and hopefully show some trends over time).

How much downtime is acceptable? In my opinion, with the type of data I am collecting, an excellent uptime percentage would be anything over 99.5 %. Anything over 98 % is good but not terrific; anything less than 90 % uptime over the course of a month is really very unacceptable in my opinion.