Ping, Latency, Jitter, and Packet Loss
Posted by on 09 December 2011 01:15 AM
VoIP services offered by ESI Hosted Services are both robust and cost-effective platforms for delivering communications to business; however, much of what makes our services so great for customers is due to the nature of the service operating over the public internet. The terms and concepts defined in this article are important components in helping assess a customers experience with our services.
(If you're looking for an easier measurement of voice quality see our article on Mean Opinion Scoring)
There are four terms which we use when measuring a customer network connection quality, which is in turn how we determine a customers voice quality potential:
PING is a standardized network utility used to test the reachability of a device on a customer network. It is also used to measure the round-trip time, typically in milliseconds (ms), of messages sent from an originating device to a destination device. PING takes advantage of the Echo Reply function built into the Internet Control Message Protocol (ICMP). As a part of normal troubleshooting, a customer may also be asked to allow PING requests (a.k.a. ICMP) through his firewall so that we can monitor customer connectivity and PING times.
Latency is the measurement of time that messages in a computer network take to arrive to a particular destination. Latency is measured either one-way (time from source to destination), or more commonly, as the Round-Trip Time (RTT) of messages and their corresponding replies from source to destination plus destination back to source. Round-trip latency is very critical to ESI Hosted Services, and VoIP in general, because it is used to determine the amount of time it will take to relay voice and call control data (aka SIP and RTP) to customer locations. Latency is almost always measured using PING and traceroute utilities including the basic utilities included with most VoIP phones, routers, and computers. For high quality Voice services it is recommended that customer network RTT Latency (PING time) average between 50-95ms during peak and off-peak usage hours.
Jitter is the variability, over time, of latency measurements between a source and destination network device. ESI Hosted Services most commonly looks for Jitter in the path between our data centers and a customer network. An important consideration is that Jitter is a somewhat imprecise term since it relies on determining a mean latency in any one situation, which can change based on a wide variety of variables. We recommend identifying Jitter by working with our support representatives and allowing time for network diagnostics to determine a mean RTT while also taking network routes into consideration.
4. Packet Loss
Messages passed between computers in a network environment are commonly referred to as "packets." Packet Loss (PL) is most commonly formed as a percentage measured variable number (i.e. 20%), and is typically calculated using PING utilities to measure the percentage of replies sent to requests. For the most popular protocols which run services like email and webpages, when replies are not received, the service initiating the request will resend the packet; however, for audio/video applications which are streamed in real-time (such as VoIP) resending packets increases overhead and is not practical since humans cannot interpret audio and video out-of-order the same way as a web browser or email client. Customer networks experiencing PL of over 2% will experience degraded call quality and a number of other VoIP service issues.
Now that you've learned the terms, here are some questions which one should consider:
How are these used?
ESI Hosted Services support will typically use the terms "PING" and "Latency" in the context of: "Your PING from our servers is higher than what we'd recommend." This translates to: "The amount of time it takes to reach your network from the servers on our network is higher than what we'd recommend." ESI Hosted Services uses PING to not only test the reachability of customer locations/devices but also measure the quality of connectivity (in terms of PING time) from customer locations.
Additionally, PING and Packet Loss may be used in order to montor internet connections and customer premise equipment in order to provide a better understanding of network quality and uptime. Typically, usage of these terms in this context may be, for example "Customer network experienced 70% Packet Loss for an hour beginning at 12am, with a complete lack of PING responses beginning at 12:30am." This translates to: "Customer internet quality became severely reduced at 12am, and finally dropped offline by 12:30am."
How do PING, Latency, Jitter and Packet Loss affect my ESI Hosted Services services?
The short answer is that like golf, the lower the score, the better one is doing. ESI Hosted Services uses these measurements to determine the timing of other protocols used for VoIP service, including voice and call control data (aka SIP and RTP). Often-times, poor measurements can be a result of network congestion, e.g., employees browsing the web, the transfer of large files, and any high internet usage activities. Additionally, certain types of internet connections are more prone to issues with these measurements than others, for instance, cable internet services use shared neighborhood routers which can cause congestion (Packet Loss and Jitter) during peak hours. For customers seeking the highest quality of voice possible, please contact your ESI Hosted Services representative to discuss internet options which include guaranteed QoS (sometimes referred to as Class of Service or CoS for short).
To measure your PING from the ESI Hosted Services network please see: Will my internet connection and network support high quality VoIP calls?