The proposition is simple: why purchase and maintain two networks when one will do?

Treating voice calls as just another form of data to be moved around the IP network slashes the cost of calls, while making call handling more flexible and intelligent.

The appeal of voice over Internet Protocol (VoIP) services is already attracting business leaders. Among UK companies, 85% have embarked on VoIP implementations or intend to do so by 2009, according to research undertaken by network operator Colt Telecom.

This is also a technology that is being driven by consumers, with the immense popularity of services such as Skype. Market watcher Continental Research predicts that home use of VoIP services will double between 2006 and 2007, from 1.8 million to 3.6 million users.

But behind these inflated expectations, there is an uncomfortable truth for the VoIP enthusiasts: sound quality – an integral part of a telephone call – is often far from satisfactory.

Vendors will argue that this is merely a teething problem, and that quality is bound to improve with time. Not so, according to a report by VoIP testing company Brix Networks which found that VoIP call quality has actually fallen since 2004, and that test users considered only 81% of calls conducted via VoIP to be of an acceptable standard of sound quality. In other tests, such a majority may seem like an endorsement of the technology, but given the near-perfect performance of traditional networks, VoIP has some ground to make up.

While there is still considerable excitement about VoIP within the business community, there is also a growing body of evidence of how implementations have fared to date. So how have those businesses that have already made the VoIP switch found the transition? Overall, the results appear to be positive: VoIP has brought cost savings, and improved the flexibility and functionality of the comms network.

If there is an overriding takeaway from the small survey, however, it is that preparation is key. Underestimating the impact of VoIP on the IT infrastructure can lead to poor quality of service and worse, can impair a company’s ability to serve its customers.

The word of law

For Michael Kolb, CIO of US law firm Dickinson Wright, the decision he took in 2003 to move telephony onto the IP network was borne of necessity. Prior to that, the firm had been using a Meridian PBX so old that spare parts were only available second hand via eBay. Purchasing a new PBX when voice-over IP seemed an inevitability would, he decided, have been pointless.

But the transition to VoIP was not quite as simple as Kolb had expected. Not only was the quality of service poor to begin with, but the main problem was voice packets being dropped: the call is essentially broken down into packets, which are routed independently across the Internet; occasionally packets go missing, leading to  sporadic interruptions in the conversation.

Soon after the launch of the VoIP system vital line-of-business services, such as itemised billing, would simply stop without warning, and the IT team would only notice once an employee had made a complaint.

The law is a business in which time on the phone is a billable resource, and the initial problems Dickinson Wright encountered with VoIP led to a dip in company profitability. “The first three months were really difficult,” recalls Kolb.

The infrastructure, it transpired, was simply overrun by the voice traffic. At first, repairing services once they had failed was a manual task that consumed hours of IT staff time just to keep the lights on. But an implementation of Compuware’s application service management suite Vantage automated the resetting of services once an interruption had been detected.

These automatic ‘eyes’ on the phone services provided the monitoring that was needed to ensure that the infrastructure could handle the VoIP traffic, and Kolb recommends any company approaching VoIP adoption to consider a similar move. “If you don’t have any kind of monitoring tool, you’re asking for trouble,” he says.

Once the application service management capabilities of Vantage were in place, Dickinson Wright was able to achieve its return on investment from the VoIP project within 6 months. Since then, the VoIP system has for Kolb been out of sight and out of mind: “All is good in my land,” he says.

Phone a friend

Vicki Williams, head of networks at insurer Friends Provident, says that the reason that its VoIP implementation did not encounter the same initial problems as experienced by Dickinson Wright was the fact that the company had tracked the development of the technology since its inception.

“Since we first heard of the idea we knew that VoIP would one day have a place here at Friends Provident,” recalls Williams. “Over 5 years, we watched it develop from a fictitious capability to an actual product with limited functionality, to an expensive product with sufficient functionality. The point when the functionality was rich and the price came down was the point at which we decided to launch VoIP.”

This meant that any network procurement and management in the years leading up to the VoIP deployment was conducted with the eventual handling of voice data in mind.

“Our data network team had been preparing for VoIP quality of service (QoS) for years in advance,” says Williams. “If you get the QoS set up and designed correctly in the first place, there shouldn’t be any difficulties when you launch.”

That said, when Friends Provident implemented its VoIP service that now supports the company’s call centre, it did take a decision to sacrifice some call quality in favour of network resources.

“There’s a choice you need to take in terms of quality, a choice between high quality and the bandwidth consumption. We found the lower level of quality to be adequate; it’s still better than a mobile phone-call,” says Williams.

How employees reacted to this dip in call quality compared to the previous PBX system depended on the nature of their work, Williams adds. “The greater the call volume and the noisier the environment the employee works in, the more important call quality becomes.”

But overall, it was not technology that was the major hurdle in Friends Provident’s VoIP implementation, but evaluating the needs of the business.

“Getting to a point where the business understood the functionality it could have, and where we understood what would have business value and what wouldn’t, was an iterative process,” says Williams.

And the business defined the terms of the VoIP implementation; it was eight months between the installation of the first phone and the last, as business units demanded changes to be frozen during periods of high call traffic.

But that careful consultation of the business – conducted in partnership with BT – ensured the close match between requirements and final results that Williams says was achieved: “We pretty much hit the nail on the head. We haven’t had to make any changes in the last two years.”

Laying foundations

Not all businesses make the staged transitions to VoIP. US-based construction vehicle manufacturer JCB, went from PBX to VoIP in one weekend.

Like Friends Provident, JCB already had the infrastructure to cope with voice traffic, including cabling that met the ‘category 5’ standard. So switching to VoIP, says JCB’s information systems manager Paul Limon, was a simple matter of installing a server and changing the phone equipment on employees’ desks. “It was a very easy implementation,” he says.

However, integrating the system with other applications, such as JCB’s SAP implementation, was more complicated. JCB wanted any of its customer records to be presented to its call handlers when that customer called. JCB hired comms maker Mitel to provide that integration, and increase the effectiveness of its VoIP system.

And while cost reduction was a driving force behind the implementation – the system had paid for itself within 16 months, eight months ahead of forecasts – there were other benefits too.

The system has made it easier for JCB to extend flexible working arrangements. One employee, while recovering from spinal surgery, was able to continue working from home, using a laptop and head set, with extended access to the virtual local area network (LAN).

Hopefully for JCB’s staff, this is something of a one-off, but the exercise proved the viability of home working, something that is likely to become increasingly critical to JCB’s business, says Limon.

“JCB as a company is growing very quickly,” he says. “Within a year or so, we will not have any room for new employees. I am positive that we will take the step of using the VoIP system to enable a large proportion of our workforce to work from home.” 

Voice of the nation

For GWR Group, which as of 2005 was merged with capital group to become GCap Media, the UK’s largest radio broadcast company, the need to modernise telephony was secondary to the transition of all of its broadcast systems to an IP platform.

But in piggy-backing the VoIP switch onto the broadcasting system refresh, the company made its return on investment case stronger, and eventually made back the money faster: indeed 75% of the return came from the VoIP project, says infrastructure manager, Aidan Hancock.

Because telephone calls to the combined company’s radio stations may be broadcast live, GCap has an unusually high demand for quality of service. For this reason, the company was fortunate to be undergoing a complete network refresh in advance of the VoIP project.

“We over provisioned our network, so that we would never have to worry about calls cutting out [live] on air, and over-specified our network monitoring tools,” explains Hancock.

Nevertheless, great care was taken to ensure that the sound quality of calls was as crisp as possible. “We did a lot testing, and a lot of fine-tuning. We got into an awful lot of nitty-gritty technical detail.”

Added to this high requirement for quality of service was some unusual requirements for call routing – calls would have to be directed reliably and instantly between studio and call centre, for example. Partnering with Cisco, says Hancock, and employing that company’s CallManager tools was vital to allow this functionality.

Ensuring user adoption at GCap was a matter of convincing senior executives and coal-face workers alike. The first to receive IP phones were the technical support team, followed by travelling executives who needed to have a single telephone number wherever in the country they were. This twin-pronged rollout helped to ignite enthusiasm for the new system, says Hancock, and it led the main body of the company – now 30 sites serving 100 radio stations – to adopt IP phones very quickly.

If there was one thing that he would do differently, says Hancock, it would be to standardise the switches between IP and PDSN – where the VoIP system meets the public phone system – on to the highest standard. GCap installed a number of switches of differing standard, and it was these that absorbed the most resources to install and support. But he adds that as public telephone operators begin to deliver IP calls, this problem will ease.

VoIP jitters

With the possible exception of Dickinson Wright, these four companies were blessed with network infrastructures that had been built with sufficient capacity to take voice traffic without disastrous consequences. And not even then did ensuring the required quality of service demand special provision.

Not all companies that have adopted VoIP are so sensible as to make these kind of provisions, says Chad Agate, president of US-based VoIP services company SIPbox, which specialises in the open source IP/PBX hybrid platform Asterisk.

“We’ve often been called in as the clean-up crew for companies experiencing jitter or echo on their IP phones,” he says. Agate says that 99% of the time these quality of service difficulties stem from an insufficient underlying infrastructure, and that this is the fault of the company’s integration partner.

Agate says he has come across many cases of integration partners using inadequate cabling, routers or switches, he says. Other companies have tried to implement VoIP over separate, departmental networks and wondered why they did not join up.

Testing and modelling is the key, he says, and that a ‘readiness test’ is the only place to start a VoIP projecy.

Dickinson Wright’s Kolb, however, believes that testing is not enough: in his company’s case, the preliminary testing that had been conducted clearly underestimated the volume of calls that the network would be required to handle.

“You can’t get any kind of test that is bullet proof, short of putting a thousand users in a lab testing the infrastructure,” says Kolb.

Prudence, then, would appear to be the key to VoIP success. Ensuring the network is ready for voice traffic, and performance monitoring once VoIP is installed helps to maintain an acceptable level of call quality.

Because while the costs may be lower in the long run, and the opportunities for creative call routing greater, if the quality of a VoIP call is poor, the bright spark who instigated the system will have some explaining to do.

Further reading in Information Age

Crosstown Traffic – July 2006

Fear and roaming in Barcelona – March 2006

BT promises VoIP war – December 2005

More stories appear in the Communications Briefing Room