Achieve High Availability in VoIP: An Implementation Example--Part I

To meet the needs for Service Availability in an open systems environment, the SA Forum specifies open standards for high availability (HA) capable of matching those of traditional voice networks. With the intent to highlight the ease of use and time-to-market benefits of a standards based high-availability (HA) platform solution, a proof of concept was created that combines SA Forum compliant middleware on a BladeCenter T platform with Asterisk, an open source VoIP solution. This article sets the stage and tells the story.

Reliability vs. availability
Reliability used to be a key metric of performance. Availability, however, goes beyond how reliable something may be and is a measure of a system or service's readiness to perform its function when it is needed. As we know, reliability in real life situations cannot be 100%, therefore highly available systems must account for less than perfect reliability using advanced methodologies, technologies and architectures.

It would seem self-evident that everyone wants reliability of both people and electronic systems. Reliability is not, however, the same as availability. Looking at a strict definition, we can say that:

• Reliability, or lack thereof, is a product or service's relative level of failure.
• Availability is a measure of a system or service's readiness to perform its function when it is needed, no matter how reliable it may be.

We can therefore see that Reliability does not equal Availability. Achieving the highest levels of availability requires an architecture and methodology that can account for less than perfect reliability. Reliability also does not cover planned down time. Availability must consider both planned and unplanned, which includes hardware and software upgrades.

Put simply, a highly available system is one that is usable when the customer needs it. Mathematically availability is defined as actual service time divided by required service time. The challenge for many of today's systems is to operate 24 hours a day, 365 days a year (sometimes referred to as 7x24, or 365x24). Expressed in percentages, availability of 99.9% for a 365x24 system has an average down time of 8.76 hours per year (525 minutes). That may not be too bad for the average PC, but it doesn't cut it for a telecommunications system. Usually referred to as X Nines, Figure 1 below, shows the scale of availability metrics and their associated downtime.

Cost of downtime
As can be seen in Figure 2, the cost of downtime across various business types is not cheap, and this does not take into account "churn" that would occur due to telecom subscribers taking their business elsewhere due to network unavailability.

Proprietary and solely home grown solutions were developed by all telecommunications systems vendors in order to meet the minimum of 5 Nines required by the service providers. The costs involved in ongoing maintenance of the software, let alone, the initial development were far from small. Open standards, such as those developed by the Service Availability Forum (SA Forum), and the products derived from them enable a significant reduction in development time. Maintenance costs for the core HA software platform are driven to virtually zero as the burden is now the responsibility of the HA middleware vendor.

The Service Availability Forum's output is a set of comprehensive specifications for the creation and usage of open High Availability middleware and systems. The specs are developed by technical working groups comprised of representatives from a broad base of industry-leading communications and computing companies. Taken directly from its website SAForum the mission of the SA Forum is to:

"Foster an ecosystem that enables the use of commercial off-the-shelf building blocks in the creation of high availability network infrastructure products, systems and services."

The benefits of adoption for network service and equipment providers as well as hardware and software vendors are clearly defined:

• Faster time to market and revenue
• Simplify deployment of new products
• Accelerate innovation
• Equipment provider choice
• Reduce product/lifecycle costs
• Increased focus on system definition, integration and value-add
• Broader technology provider choice, pick from "best in breed"
• Larger market for standard products
• Reduced need for product customization
• Simplified ecosystems for partnering and more integrated solutions