Interpreting the Results of the HostMyCalls Packet Loss and Delay Test
The HostMyCalls Packet Loss and Delay Test Tool (Tool) uses Internet Control Message Protocol (ICMP) with low Time To Live (TTL) values to find and get ping responses from each router or hop in a network. This method allows the Tool to detect route changes as they happen. If the IP address you entered experiences route changes during your specified data collection time window, each unique route is assigned a Route Pattern ID (RP). Similar routes are grouped together under the same RP. Statistics and reports are grouped for each identified RP during the data collection process.
Ping packets are sent to each IP address in the route every 3 to 4 seconds. Ping packets not returned within 1500ms or 1.5 seconds are considered lost. Statistical data is updated in the database and available for the browser based reporting system every minute.
This tool can monitor the IP address you specified for up to 7 days. It is recommended that you monitor for at least 12 hours with 6 of those hours during normal business operation before you draw any conclusions from these reports.
Some Internet Service Providers (ISPs) program their routers to discard packets with low TTL values which may render these reports incomplete. Hops or IP addresses may not be identified for this reason. In addition, some Internet providers share the traffic among several routers. Two or more IP addresses will be listed for the hop in these instances.
In viewing the graph on delay times, several hops may have response times greater than the next hop in the route. Forwarding a packet to its next destination takes less CPU time than responding to a TTL expiration which accounts for this apparent anomaly.
Although packet loss and delay may be caused by the same problem, treat each issue separately as you read the reports.
Finding the Source of Packet Loss
The IP Address that was monitored is referred as the Destination IP. We believe a quality Internet connection should have 0.1% packet loss or less. If the packet loss is 0.1% or less at the Destination IP address, then packet loss at any of the other hops can be ignored.
If packet loss is greater than 0.1% at the Destination IP address, examine the hops immediately ahead. Do they exhibit similar loss percentages? If they do, then the first hop where lost packets start occurring could be the cause.
Examine the timeline graph at the bottom of the RP report page. If the lost packets are grouped together during times of the day, narrow the report period at the bottom of the page just before the beginning and just after the end of one of the groups of losses. Does the loss of the suspect hop still exhibit similar loss percentage? If so, click on the hyperlink of the address to compare the two hops more closely. Do the losses over time still seem similar? If not, consider examining another hop as the cause of the packet loss.
Source of Packet Loss at Destination IP Only
If packet loss is occurring greater than 0.1% at the Destination IP and not the hops immediately preceding, then the source of trouble is between the last hop with insignificant packet loss and the Destination IP or in the device responding to pings for the Destination IP.
Some factors that can cause packet loss at the Destination IP include:
1. Equipment problems with the cable/DSL modem or router including potentially over-heating equipment.
2. Undetectable bridge/switch or other networking hardware (i.e. hub) in front of the Destination IP device.
3. Congested Internet connection. In other words, too much traffic for the delivered speed. Please note that delivered speed may be different than the expected or promised speed.
4. Wiring outside or inside the building but before the device which responds to pings for the Destination IP. Quality of wiring is poor or deteriorated from weather, age, animals and other environmental factors.
5. Environmental interference from microwaves, wireless radios, refrigerators, etc too close to the wiring or cable/DSL modem or router.
If your Internet connection is a DSL circuit:
1. Other circuits, such as commonly used T1s, in the Local Exchange Carrier’s (LECs) cable bundle could be interfering with the DSL circuit’s frequency.
2. The DSL splitter, which divides the phone line from the DSL circuit, could be faulty.
3. The LEC could have load coils or bridge taps on the cable which is delivering the DSL circuit.
4. Other analog devices like fax machines or cordless phones have been attached to the phone line before the DSL splitter splits off the DSL circuit.
If the Internet connection is supplied by cable modem:
1. The coax cable passes through more than one splitter before the cable modem.
2. Cable modem attached to wrong port of the splitter.
3. The decibel level of the cable TV transmission is too low. The cable repairman will frequently check this first. Be advised that the decibel level can change throughout the day.
Finding the Source of Latency
In the AvgDelay/ms column look for hops where the delay increases significantly compared to the other hops and the delay seems to affect all other hops greater in value. An example would be a report where the average delay jumps from 15ms to 35ms from hop 14 to 15. All hops 16 and above are approximately 35ms or greater. This means a significant delay is induced between hops 14 and 15. Because hops 16 and above stay in the 35ms or greater range, it suggests the increase delay from hops 14 to 15 was not caused by the router’s response to a ping packet. If one of the problems with the Internet connection is delay or latency, then this could be a contributing source.
If you suspect a hop is inducing delay on the destination IP address, click on the hop’s IP address link to compare its delay over time with the delay on the destination address. The time range of this report can be narrowed down to as fine a granularity as 16 minutes in 1 minute segments. Look at the line graph of each IP address over time. When the suspect problematic hop increases its delay at a certain point in time, does the destination IP address also have an increase in delay? If so, then the report definitely suggests an impact on the Destination IP by that hop or one earlier that is affecting both hops.
If there is a large increase in delay only at the Destination IP, then review the sources of packet loss at Destination IP above.
Run the Report Again
Once a potential cause of trouble has been eliminated, run the report again. Is there a difference? If not, look for another potential source of trouble. And do not forget, many times there can be multiple sources inducing problems on an Internet connection.
Prior to problems, run baseline reports. The reports should cover at least one week starting 1 hour prior and ending 1 hour after the normal business day. For the best results the report should not cover holidays or other times when your office is closed. Additionally, run baseline reports on regular high traffic days for your business such as the day after a holiday for technical support company or the week prior to Valentines Day for florist, restaurants, etc.
Comparing the baseline reports to reports when there are problems can help identify problem areas faster, minimizing repair times.