Holistic VoIP lifecycle management

Enterprise Voice over IP (VoIP) adoption continues to grow exponentially. A 2005 study by International Data Corp. (IDC) projects business VoIP will hit $7.6 billion by 2008, a compound annual growth rate of 282%.

VoIP provides a true challenge for most organizations because it combines the worlds of traditional voice managers who talk in MOUs and POTS with data managers who think about Mbps and packets. With VoIP, the paradigm shifts for the voice group by changing standards from number of trunks needed to number of concurrent calls supported. For data managers, there is a huge difference between an e-mail that takes 30 seconds to transmit as opposed to when a sales manager picks up a phone to call a customer and there is no dial tone.

The term holistic lifecycle management sounds like an academic description developed by a group of analysts sitting for days on end in a windowless room. While the term may sound academic, the process is a tested, straightforward approach enterprises need to follow to successfully roll out and manage VoIP. A holistic approach to VoIP focuses not only looking at VoIP in and of itself but how VoIP affects other applications and users across your infrastructure. Lifecycle management focuses on the evolving nature of your network infrastructure. What happens tomorrow will likely be very different from what happened today.

• Assess: Determine network readiness
• Monitor: Discover and isolate performance issues
• Manage: Troubleshoot and resolve issues
• Validate: Observe performance by call
• Report: Baseline and trend
• Control: Improve VoIP performance

As we delve deeper into each of these six steps, you will notice there is some overlap between each one. This is to be expected, as there is no clean line of demarcation between each process. Holistic lifecycle management is an ongoing, dynamic process intended to improve the performance of a VoIP deployment while maintaining existing data application integrity. The ongoing nature of lifecycle management assumes the performance of VoIP and all other applications will never be perfect – however, it can be continuously improved over time.

Determine Network Readiness

To have a successful VoIP rollout, the voice traffic must coexist peacefully with existing business-critical data traffic. Without a network assessment, your enterprise runs the risk of major performance issues – both with existing applications and with the VoIP rollout. Too often, enterprises assume they have adequate bandwidth and then must firefight major problems during implementation. Or enterprises add more bandwidth at every location assuming additional bandwidth requirements will be consistent network-wide.

There are three ways to complete a network assessment for VoIP: synthetic call transactions, management information base (MIB) polling and granular user data via passive monitoring. The synthetic call approach places simulated calls across the network and characterizes the quality of the calls. While this approach can help quantify parameters before deployment, including simulated Mean Opinion Score (MOS), the synthetic calls do not provide granular analysis of how existing data traffic impacts VoIP availability and quality. In addition, it is actually adding management overhead to your existing critical application traffic.

As a second option, many enterprises conduct an assessment using a reporting tool that polls standard MIB data from routers and equipment. This approach leverages traffic data in the analysis but typically averages the utilization across five-, ten- or fifteen-minute buckets. The weighted average can easily skew data so a network that cannot adequately handle a VoIP deployment appears to be ready for VoIP calls. This false sense of security can actually be worse than not doing an assessment at all.

The third approach for a network assessment relies on actual user data from across the enterprise to determine VoIP readiness. The key is passive monitoring that provides up-to-the-second granularity. By measuring actual user data, enterprises can determine how many simultaneous VoIP calls at certain quality levels can be handled on the existing network without negatively impacting existing data applications. Without granular user data, you are taking an educated guess on the impact of a VoIP deployment. The challenge for many enterprises is how to collect granular performance data without impacting the network itself. Detailed and granular user data takes the guesswork out of assessing network readiness, so you now know where additional resources will be required before the deployment begins.

Find and Isolate Performance Issues

For most enterprises, the acceptable end-user experience varies substantially from traditional voice to data applications. Voice managers historically have little or no tolerance for variations in call quality while data managers are most accustomed to data’s traditional best effort philosophy.

Once VoIP is deployed, monitoring detailed actual traffic – both voice and data – is the next step to identify and isolate performance issues. Trouble tickets will flood network managers when poor VoIP quality is noticed. A huge challenge for most enterprises is the perception of VoIP. Studies have shown that users believe quality is much worse for VoIP than traditional PSTNs – even though actual measurements prove this is misconceived. What this means for network managers is that a VoIP deployment will likely increase the number of performance complaints, no matter the actual call quality. This can waste an IT department’s valuable time trying to track issues that, from their point of view, don’t exist.

Monitoring actual VoIP usage in conjunction with other data traffic helps eliminate the guesswork in performance monitoring. With proactive monitoring, enterprises can determine the true cause of poor VoIP performance – whether it be new applications, network issues or service level parameters. With enhanced visibility, enterprises can also isolate performance issues by pinpointing the cause of the performance degradation that impacts end-users.

Troubleshoot and Resolve Issues

Following closely behind monitoring – and many times in tandem with that important step – is troubleshooting and resolving problems associated with poor performance. As mentioned earlier, network managers typically must respond immediately to complaints about no dial tone or unacceptable call quality. There are many potential causes of poor VoIP performance. These range from physical problems on the local loop to an over-utilized port to misconfigured class of service (CoS) settings to high levels of jitter within the voice application itself.

Once the cause of degradation is isolated, troubleshooting the problem can greatly reduce the amount of time to resolve the issue. While a problem is occurring, seconds and minutes saved often go straight to the bottom line.

A bigger challenge than real-time issues for many enterprises are the intermittent issues that come and go. Intermittent issues typically grow and cause more havoc over time. These occur when you least expect it. When it comes to VoIP, it is even more critical to find the pesky intermittent problems and resolve them before they grow and more users are negatively impacted. Network managers must be able to troubleshoot issues ranging from the local loop to the port to service level parameters across each and every site.

Performance by Call

As discussed earlier, one of the biggest challenges for network managers is validating performance by VoIP call. MOS has become a standard for gauging the quality of individual call performance. In order to understand the end-user perspective, network managers must dig deeper than just MOS to validate what impacts the performance of each call.

Too often, many enterprises make a critical mistake in averaging the call performance. For example, the IP PBX may provide statistics on average MOS scores across a site, but an enterprise may be at risk when they feel they exceeded a target MOS score 99% of the time. The key is validating performance to optimize the VoIP performance to improve the other 1% of unacceptable calls.

Holistic lifecycle management intertwines different components to optimize the VoIP application performance. For example, it is not adequate to just know a call may have a low MOS, it becomes imperative to know the true cause of the poor performance – whether it be jitter, latency, a layer 1 issue or another application fighting for critical resources. Enterprises need up-to-the-minute access to this information to validate performance across each of the potential causes of poor quality.

Baseline and Trend

A wise adage for network managers to follow is that good organizations baseline performance, but great ones continually baseline and manage performance. In today’s world of new applications, remote users, security threats, and mergers and acquisitions, the importance of reporting becomes more critical. What has worked well over the past year may no longer be sufficient when deploying VoIP or other peer-to-peer applications on your infrastructure.

With strong reporting capabilities, you can be much more proactive in optimizing your application and network performance. For example, instead of waiting for a problem to occur – such as over-utilization negatively impacting individual call performance – and then reacting to the situation, trending could identify the growth of utilization over time. Thereby you can add bandwidth where needed before end-users are impacted.

Most enterprises do not remain stagnant when it comes to networking requirements, so a one-time VoIP assessment is soon outdated. You may have a merger, add a remote office or deploy another application. Each and every one of these changes can impact the quality of VoIP performance. That is why trending is so valuable for your organization. What may work fine today for a location may be inadequate in six months. By reporting and trending, you can plan for the new requirements over time instead of firefighting issues after they have occurred.

Improve VoIP Performance

The final step in the lifecycle management approach is to leverage the first five steps in order to proactively control your VoIP application and network to improve overall performance. The control parameter may encompass many different areas depending on where quality may be improved in your organization. Control can include aspects such as:

• Increasing bandwidth to handle additional usage caused by VoIP
• Leveraging class of service capabilities with an MPLS-deployment
• Improving service level parameters from the service provider
• Shaping traffic so the most business-critical and delay-sensitive applications have priority
• Eliminating recreational applications such as file sharing and streaming media

The control parameter requires a holistic approach to performance management. Simply looking at a single application, site or user where a problem may be occurring is not enough. It becomes imperative to not only look at that individual issue, but evaluate how that issue might impact or be impacted by other business-critical applications throughout your infrastructure.

By approaching VoIP implementation and management in a holistic manner, you can ensure you are ready for a VoIP deployment, monitor its progress, proactively manage and troubleshoot issues, validate call quality, report on deployment success and control parameters to improve performance. As your network and applications evolve, you will be ready for the new challenges in managing performance across your entire infrastructure.

Get the complete version of this whitepaper by visiting: http://www.visualnetworks.com/resources/WP_voip2005_F.pdf

Learn more about Fluke Networks’ VoIP solutions at: http://www.flukenetworks.com/us/Solutions/VoIP/VoIP+Lifecycle+Management+Solutions.htm