To ensure a clear understanding of the terminology used in this documentation, refer to the terms and definitions listed below:
This is the time it takes to move voice packets from the phone to 8x8 (and back). In general, this should not exceed 150 ms in one direction to prevent deterioration of voice quality.
Note: When part of the call travels over the public Internet (which introduces its own latency), the organization's internal network latency should be significantly less than 100 ms.
Compensating for latency
In most cases, latency is introduced by a congested network or a carrier issue. 8x8 suggests that you run PingPlotter, a free 3rd party utility, to get a better overall view of your network as well as a bufferbloat test.
Jitter is a variation in packet transit delay caused by queuing, contention, and serialization effects on the path through the network. In general, higher levels of jitter are more likely to occur on either slow or heavily congested links. These are the variations in packet delay. As far as the source endpoint is concerned, the packets would be sent in a continuous stream. But since each packet may take a different route to its destination, network congestion or improper configuration can result in variations in packet delay. If the packets are not received in the same order or were dropped entirely on the way, this results in Jitter. Jitter that exceeds 40ms will cause severe voice quality issues. High levels of jitter are usually a consequence of slow speeds or congested networks.
Jitter may be measured in a number of different ways, several of which are detailed in various IETF standards for RTP, such as RFC 3550 and RFC 3611. Some of these methods are Mean packet to packet delay variation, Mean absolute packet delay variation, Packet delay variation histograms, and Y.1541 IPDV Parameter.
Compensating for Jitter
The first thing to check is to verify the QoS settings of the company network. It is expected that the increasing use of "QoS" control mechanisms such as class-based queuing, bandwidth reservation, and usage of higher-speed links. If the QoS has not been configured or improperly set up, voice packets will not be receiving the proper priority. This will result in missed or discarded packets. Audio calls will then be subjected to high levels of jitter, degrading the voice quality. If the QoS settings are correct and network traffic is at its usual levels, there should not be any significant jitter.
Jitter buffers are not recommended, Although the size can be increased up to a point. Generally, they are only effective for delay variations of less than 100 ms, and even then, deterioration in quality may be easily noticeable to users. There are various tools available that can be used by experts to isolate jitter sources and remove them. Therefore organizations should try to identify the sources of jitter on their networks instead of relying on jitter buffers.
Fragmentation is a technique that divides a data packet into smaller data packets so that they can be sent through a network that can only transfer small data packets. Fragmentation occurs during network transmission. When these packets are received at their destination, they are reassembled to their original data packet size.
In rare situations, signaling packets may become extremely large and require to be fragmented. If fragmentation is not supported, there will be a loss of audio. 8x8 recommends that fragmentation be supported by all network equipment, including routers.
Packet loss is caused within routers. Errors in routers can and will occur if a router is overloaded. This will result in the router dropping packets; in IP telephony, Packet loss is unacceptable. The performance of an IP call will suffer if a packet loss reaches more than 0.3%. Packet delays can wreak havoc in latency-sensitive applications such as VoIP. Delays can be introduced in the data network due to router configuration, network capacity and performance, and load on the equipment.