Continuity planning for VoIP

Lawrence Imeish provides advice on how to harden and protect VoIP and IP telephony networks.

As more organisations look to IP telephony and VoIP to boost productivity and reduce overheads, business continuity managers should re-evaluate their existing data networks to assess the new challenges faced. Adding VoIP or IP telephony can be a relatively simple, or very difficult process. Any VoIP integration specialist will tell you that spending time planning the service, will produce a stable and readily accepted system. Adding voice to a data network magnifies the importance and risks of the network. The integrated network becomes an extremely critical piece of the business and should be well protected. Imagine the loss in productivity and revenue when a network outage affects not only the users’ ability to access network applications, but their ability to make and receive calls.

Business continuity planning for IP telephony involves examining every layer of the network and all its dependant components. The most important part of any IP voice system is of course the IP. When adding critical services such as voice to an IP network, be sure to understand the service is only as good as the network that carries it. Many times VoIP and IP telephony services fail due to lack of attention to basic IP connectivity and IP QoS.

IP network link redundancy becomes vital when supporting an IP telephony system. Most IP voice systems require centralised call processing; this is where the call routing decisions are made based on dialled digits, time of day, and network utilisation, etc. Loss of connection to the call routing service or application failure would lead to all IP phones no longer able to make or receive any calls. For this reason, all WAN links that carry voice or voice signalling should be made redundant with shadowed circuits or at the very least ISDN backups. All backup links should be tested on a regular basis during maintenance windows to confirm they automatically cut over when the primary link fails. The IP routing protocols (OSPF, EIGRP) should also be optimised for fastest possible convergence upon a network change and should recognise VoIP as ‘interesting traffic’ capable of initiating the backup link. It’s beneficial to test the failover configuration and measure the time it takes for calls to pass after a WAN Link failure.

In terms of IP and Ethernet redundancy, it’s best that each edge switch have at least two links to the core switches. This will create redundancy in the LAN even if one of the core switches fails. It is difficult and expensive to cluster the edge switches, a secondary solution is to have a cold unconfigured standby switch ready to deploy in the event the edge switch fails.

VoIP and IP telephony systems can terminate calls in one of two places; to another IP telephony device/endpoint or a VoIP gateway that interfaces to the local PSTN. Call routing redundancy is often overlooked when deploying or managing IP voice systems. If the enterprise has multiple offices or multiple gateways, backup call routing schemes should be devised so that outbound calls will make it to the PSTN even if the preferred local gateway or PSTN service is down. For example if there are two offices, one in New York and one in Washington, and the local Washington gateway fails, all the users outbound calls from Washington could be sent to the NY via a WAN connection (leased line or internet VPN) and dialled back to Washington using the NY gateway. This could be an expensive solution because during the local outage, all outbound Washington calls would incur long distance charges and the backup gateway would need to be designed to handle the extra call capacity. Another backup option is to send the organisation’s calls to an Internet telephony service provider for termination; usually $0.02 - $0.05/min for North American calls. This creates new security concerns but is a viable option for companies that do not want to invest in the additional hardware costs. A correctly designed and managed redundant IP Voice system will take advantage of IP’s inherent survivability/rerouting capabilities. This, plus VoIP’s ability to make call routing decisions contingent on the IP network’s status, makes for a robust and survivable voice communications system.

Each vendor has their solution for a high availability call routing service, usually a central cluster or two, or nodes with decentralised limited call routing services as backups (i.e. Cisco Call Manager and Call Manager Express). When deploying voice services, take advantage of the vendors’ clustering and redundancy solutions and integrate them in the IP voice network, it will pay off exponentially in the event of a hardware or application failure or outage.

An important IP voice design and management detail is the need to ensure that every network element that carries IP voice has adequate UPS and backup power. This extends to more than just the routers, switches and gateways, to include any DSX panels, call managers, CSU/DSUs, Wireless bridges, and even locally powered IP telephones. An effective design for LAN IP phone redundancy is to use edge switch to power the IP phones, this saves the need for a separate UPS for every phone set. If the switch and phone support 802.3af (Power over Ethernet) it would be wise to make use of it. It is helpful to create a flowchart that traces a typical IP call and note each powered device at every ISO Layer that the RTP Stream (the call) or call signalling traverse during the course of a call, end point to end point and confirm they all have proper backup power. If the power goes out to the office during an emergency the IP phone may be the only means of calling for help, therefore it is crucial that the phone and the whole IP voice system always function.

Finally security in an IP voice environment is also magnified; a hacker in the network will not only steal or damage files, but could possibly access the VoIP/PSTN gateways and prevent legitimate users from making or receiving calls or even make thousands of dollars in unauthorised calls. IP voice security is an extension of basic IP security; it is assumed the network already has existing firewalls and other security architecture. If it does not, then IP voice should not be deployed until the IP security is hardened. IP voice has special requirements for firewalls and user authentication. Depending on the protocol the system uses - H.323, SIP, Skinny - the firewall may need to be configured to allow incoming RTP traffic over a range of UDP ports, this can create a security risk. The best solution is to use VPNs between sites connected through the Internet and to ensure that those VPN devices have the processor power required to pass the voice packets with minimal delay. Another solution is the newer VoIP specific firewalls that actually work with your call routing systems to dynamically open and close ports on a per call basis. Needless to say any calls (or data) crossing the public internet should be encrypted. IP voice also allows for real time call accounting and active toll fraud detection, these features should be activated if they came with the billing software

Although it may appear that VoIP creates a more complicated and vulnerable network than traditional TDM based telephony, it is important to point out its benefits. The strongest factor in deploying IP telephony is the direct monthly cost savings reaped from a substantially reduced telecoms bill. Enterprises have seen their phone bills slashed by as much as 60 percent when deploying IP voice. There are other factors such as improved worker productivity, faster customer response times, and reduced cost of any moves, additions, or changes due to web based user provisioning. The deployment of IP voice services also provides an opportunity to unify your network and combine what were once two separate systems into one manageable, integrated, redundant unit. Finally IP telephony sets the stage for new applications and services. Imagine having the power to access vital company data and applications directly from an IP phone or a Wi-Fi IP phone handset while making/receiving calls, or receiving your voicemail via e-mail and on a web page. These new applications and services will make it very difficult for anyone who has experienced IP voice to go back to legacy PSTN. In fact, a properly planned and managed IP telephony or VoIP network will give the enterprise as much if not more survivability and availability as a comparable legacy system and still provide new features and applications which the legacy system simply cannot.

Lawrence Imeish is a consultant specialising in IP telephony and VoIP deployment. He holds a CCIE, and has over 10 years planning, deploying and managing enterprise networks. He can be reached at LawrenceIme@hotmail.com