High availability represented by clustering services has become increasingly essential for organizations deploying business-critical VoIP applications.

A cluster is a group of computers working together to run a common set of applications and to show a unified system to the client and application. The computers can be physically connected by cables (or via a Local Area Network or Storage Area Network) and programmatically connected by cluster software. These connections allow computers to use failover and load balancing, which is not possible with stand-alone computers.

Clustering means having several machines sharing the load at different rates, so administrators need to do clustering in order to reach high availability. Failover is an intrinsic feature of clustering. This means that if one single node is down, the whole cluster continues to be in service. There are several machines acting together and if a part of the system is down, another part is still in production, which gives a chance for technicians to go and troubleshoot the system. It’s also used for upgrading software, for instance, an administrator can just upgrade their software node by node without taking down the whole system.

Clustering is not a new solution. Clients must have used clusters to compensate for not having enough space on their hard drives in the late 1950’s to early 1960’s. In 1967, parallel computing was invented by Gene Amdahl of IBM. Amdahl's Law describes mathematically the speedup one can expect from parallelizing any given otherwise serially performed task on any parallel architecture. Amdahl wrote an important article which defined the engineering basis for both multiprocessor computing and cluster computing, where the primary differentiator is whether or not the interprocessor communications are supported "inside" the computer (on for example a customized internal communications bus or network) or "outside" the computer on a commodity network.

One can have clusters of CPUs (hardware) and databases (software). For telecommunications architecture needs, clustering can be done in order to bring high availability to some of the more sensitive parts of the architecture, mainly the management system (which includes billing and user profiles), and the switching systems. This is always a key issue when it comes to VoIP server maintenance.

Our failover solution is based completely on configuring the Microsoft Windows 2003 Server network load balancing; however, the database uses a different technology to provide high availability: the Microsoft database replication mechanism. Thus the database located on the master machine instantly replicates itself on the backup machine. After an eventual failover, the backup machine will be able to process calls using fresh data. Without Dialexia’s failover configuration all data will be lost in the event of a catastrophic failure.

We suggest that you install your server and gateway in a redundant configuration, with no single point of failure and make use of Static Load Balancing. Set Dial-Gate® to fast failover mode, which we will discuss in more detail, below:

The availability of Dial-Gate® could be increased by putting redundancy components and network protection in place. A real increase in availability could be achieved by running two servers in Redundancy Mode. Dial-Gate® has tested well with two different Redundancy Type “1+1” scenarios. Redundancy Type “1+1” means that one server is used to serve all customer interactions and that one server is used in Hot-Standby Mode. In a situation where the primary server fails, the hot-standby server will take the full load.

Entry Level Redundancy Servers:

As we mentioned earlier, Entry Level Redundancy configuration is based on ‘Windows Network Load Balancing’ or usually referred to as ‘NLB’. Setting Dial-Gate® in Redundancy Mode based on NLB will allow an automatic failover - meaning that the hot-standby server will take the load when the primary server is declared dead. In a NLB configuration, the hot-standby server will be able to render most system functionality. To return operations to the primary server and to allow full functionality to be served, an action will have to be taken by the administrator.

Enhanced Level Redundancy Servers:

Enhanced Level Redundancy configuration is based on ‘Windows Clustering’. Setting Dial-Gate® to Redundancy Mode based on Windows Clustering will allow automatic failover and automatic fall back. The main task of the administrator is to resolve the issue that renders one of the servers non-operational. To allow Dial-Gate® to run on a Windows Cluster system, you will need to make use of a SAN and setting MS-SQL to Cluster Mode will be required. During the switch-over period, service will not be available for a period of 30 seconds (this value is server-dependent). Contact your Windows Cluster specialist for proper configuration and operation of these components.

Here is a comparative table:

Both Dial-Office and Dial-Gate offer redundancy using high availability. Failover offers a solution in mission critical situations when a server is moved offline intentionally for servicing, or when it runs into problems. With high availability, the VoIP server administrator can rest easy knowing that their system will keep running without any obvious changes on the client side when one server goes down.