/mcpNote: No fragmentation is required if the fragment size is larger than the link MTU size. For example, for a T1 link with a 1500−byte MTU, the fragment size is 1920 bytes. Therefore, no fragmentation is required. The packet fragmentation size should never be lower than the VoIP packet size. Do not fragment VoIP packets. Fragmenting these packets causes numerous call setup and quality issues. There are currently three link fragmentation and interleaving mechanisms available. For further explanation of various delays introduced in a packet network, refer to Understanding Delay in Packet Voice Networks. This table lists their benefits and limitations: Link Fragmentation and Interleaving (LFI) Mechanism Description Benefits Limitations MTU fragmentation with WFQ Interface level command to change MTU size or IP MTU size. Used to fragment large IP packets to specified MTU size. LFI uses WFQ to interleave real−time packets in between the fragments. Simple configuration. Fragments are reassemble only by the receiving application. Therefore, inefficient use of the network. Only IP packets with the Do not Fragment (DF) bit not set can handle fragmentation well. Highly processor intensive. Not recommended. Multilink Point−to−Point Protocol (MLPPP) LFI On point−to−point serial links, MLPPP must first be configured, then a fragmentation size must be set in ms. Interleaving must also be enabled on the multilink interface. Packets are fragmented on one end of the link and reassembled at the other. Several links can be combined to act as a large virtual pipe. Only available on links configured for PPP. Solutions for PPP over Frame Relay or PPP over ATM are also supported in Cisco IOS Software Release 12.1(5)T or later. Frame Relay Fragmentation (FRF.12) On Frame Relay PVCs, the frame−relay traffic−shaping command must be enabled and a Packets are fragmented on one end of the PVC and reassembled at the other. Only available on Frame Relay PVCs with the frame−relay traffic−shaping command