Managed Services
Plan ahead for branch office VoIP
Enterprises with remote offices should consider a variety of factors when choosing a service provider.
by Gregory Davis
Businesses with multiple branch offices can realize substantial savings and productivity gains by upgrading from a traditional PBX or key system to an IP PBX, but the network should be secure and deliver the bandwidth and performance required to handle voice-over-IP (VoIP) traffic.
While most offices will require a T-1 or more to accommodate voice and data traffic, small branch offices and retail locations may be able to save with business-class DSL if they are not using much bandwidth for data applications. Symmetric DSL (SDSL) provides the same amount of bandwidth in both directions and is more consistent in its performance characteristics than asymmetric DSL (ADSL). The service-level agreements (SLAs) for ADSL services have lower performance and reliability thresholds than those for SDSL services. Likewise, the SLAs for SDSL are weaker than those for T-1s.
Businesses need to consider what the impact will be if there is an outage or performance degradation that impacts call quality at a particular site. This may not be a significant issue for offices with few employees, part-time staffing or personnel who regularly use cell phones. Larger offices and headquarters locations demand maximum uptime and have bandwidth requirements that necessitate one or more T-1s or a DS3 connection.
Service providers offer a wide selection of access technologies in the cities where offices are located, or even nationwide so they can continue to support growing businesses. Most service providers, including big name phone companies, only have regional coverage using their own last-mile facilities. Service providers can often cost-effectively augment their own service areas with other providers' DSL and T-1 access to offer nationwide coverage.
When connecting branch offices for VoIP, the service provider should connect remote sites via a private network, instead of over the public Internet. Since VoIP is performance sensitive, running it over the Internet can lead to poor sound quality and dropped calls. Connectivity over the Internet is inconsistent and can suffer from packet loss (lost data), latency (delay) and jitter (variable delay). This may not be an issue for other applications, since lost or delayed data can be resent and users may not even notice delays.
VoIP is a real-time application, however, so all packets need to get through quickly and in the same order they were sent to prevent poor sound quality or dropped calls. In addition, private T-1s and SDSLs enable service providers to prioritize VoIP traffic in both directions, to and from each site, for end-to-end quality of service.
Since VoIP is more performance-sensitive than other applications, the service provider should prioritize it over other traffic. The only traffic that might take a higher priority than VoIP is real-time financial transactions. Most service providers prioritize VoIP traffic using a routing technology called multiprotocol label switching (MPLS). MPLS allows the creation of multiple classes of service, prioritizing and reserving bandwidth for different types of traffic.
Service providers typically offer up to five classes of service:
- real-time 1 for VoIP traffic;
- real-time 2 for videoconferencing traffic;
- data 1 for mission-critical data, like financial transactions and credit card data;
- data 2 for enterprise applications like SAP, Oracle and video surveillance; and
- data 3 for low-priority traffic such as Web browsing and FTP.
Classes of service can be assigned based on application or source/destination IP address, which is useful if all high-priority traffic originates from a specific LAN segment, as may be the case if there is a separate Ethernet drop in each office for VoIP. This approach requires that the customer premise equipment (CPE) router that the service provider uses be configured to manage the class of service prioritization.
If traffic is prioritized based on application, class of service can be managed on the CPE by the service provider or at the application servers by the customer. For customer-managed class of service, the customer's application servers mark the traffic based on the appropriate class of service for respective applications. This requires some level of expertise to ensure it is done correctly and circuits are not over-subscribed with high-priority traffic.
The IP PBX needs to be secure and VoIP traffic kept private. Keeping VoIP traffic off the public Internet by using private DSL and T-1/DS3 circuits avoids the use of public IP addresses that can be directly attacked by hackers or subject to denial-of-service attacks. Appropriate firewall, intrusion prevention and antivirus solutions, whether through the service provider or self maintained, should always be in place on the network to block hackers, viruses and other malicious attacks.
Gregory Davis is vice president of product marketing at
MegaPath, Costa Mesa, Calif.
For more information
(click here)