Voice is sensitive to delay, jitter, and packet drops. What are the recommended maximum values for each of these?

High-quality voice and interactive video have the following network requirements:

• A maximum of 150 ms of one-way delay
• A maximum of 30 ms jitter
• A maximum of 1 percent packet loss

When you use the Modular QoS CLI, or MQC, what steps are involved in setting a bandwidth limit for voice traffic that is sent out an interface?

Data packets can be large compared to voice packets. Why is this a problem across a WAN link, and how can you remedy it?

Although voice might be placed in a priority queue, a voice packet can be delayed by a larger data packet if it arrives at the interface after the data packet has begun being serialized. To maintain the delay budget for voice, you should serialize all packets in 10 ms or less. This is not a problem on links of T1 speed or better, because you can serialize a 1500-byte packet within 10 ms. To remedy this, you can use LFI on slower-speed links.

What is the difference between fax/modem relay and passthrough?

In relay, the fax or modem analog data is demodulated by the sending gateway, packetized, and sent over the IP network. The receiving gateway remodulates it and forwards it as analog data to the fax or modem.

In passthrough, fax and modem calls are treated as any other analog voice call, with the data carried in-band in RTP packets to the remote fax or modem.

What are the two types of fax relay that Cisco routers use, and which is the default type?

Cisco routers use Cisco proprietary fax relay and the ITU-T standardized T.38 fax relay. Cisco fax relay is the default.

In which configuration mode are fax/modem commands given on MGCP gateways? How about H.323 and SIP gateways?

Fax/modem relay and passthrough commands are given at global configuration mode on MGCP gateways. On H.323 and SIP gateways, they can be given at either dial peer or voice service configuration mode.

How does SRTP protect voice media traffic?

SRTP encrypts the RTP voice media payload, but not the RTP header, using AES encryption. It authenticates the RTP header and payload contents by computing a one-way HMAC-SHA1 hash and placing the results in an authentication tag at the end of the packet. The receiver runs the same computation and compares its result to the contents of the authentication tag. If the contents do not match, the receiver drops the packet. SRTP also includes a replay protection process to avoid DoS attacks.

When using encrypted voice within a LAN, why is it a good idea to also encrypt traffic between the voice gateway and Cisco CallManager?

Voice media and signaling are not the only types of voice traffic that traverse the WAN. MGCP gateways communicate with Cisco CallManagers. IP phones download TFTP files and Dynamic Host Configuration Protocol (DHCP) information, if those servers are centrally located. DTMF tones and encryption keys might be exchanged. If you are encrypting the IP phone traffic, it makes sense to encrypt the other voice traffic also, unless your network is secure and trusted.

How is firewall function affected if an IPsec tunnel from a remote gateway terminates on the Cisco CallManager, rather than another device?

A firewall cannot thoroughly inspect IPsec traffic coming from the WAN and going through the firewall into the LAN. All the firewall sees is the IPsec header. It would have to be able to decrypt the packet to see the original headers and the payload information. Thus, terminating an IPSec connection on the CallManager prevents the firewall from inspecting that communication, but it also secures that traffic while it traverses the network to reach the CallManager.