As more and more enterprises are transitioning their voice, data and video services onto a single network, the need for IP lifecycle management tools and practices is more critical than ever. Whether...
September 1, 2006
As more and more enterprises are transitioning their voice, data and video services onto a single network, the need for IP lifecycle management tools and practices is more critical than ever. Whether looking at voice or data delivery, the issues that potentially threaten performance are pretty much the same across the board. The only difference is, there is a lot more riding on a single network, which means absolutely no margin for error in performance optimization.
The “everything over IP” approach offers countless administrative advantages. It allows enterprises to cut costs, increase efficiencies, and centralize network administration and control — a boon for meeting today’s security requirements.
With a properly designed and maintained infrastructure, the first order of the day for network managers is to put any concerns to rest during planning, deployment and ongoing maintenance and support.
If the IP network is properly designed and tested, sound, data and video quality will have no problem meeting end user expectations.
In the case of VoIP for example, the quality of sound over a well-performing network can be so clear, some IT managers actually tune phone systems to add slight static so it sounds more like a traditional phone call.
While there is no doubt that collapsing everything onto a single IP backbone generates significant business benefits, there is still a great deal of misunderstanding about the challenges posed by this approach.
The perception is that it will make everything simpler. As long as IT managers can measure the basics, such as latency (delays), packet loss, and outages, they think the network will be just fine.
However, voice and other real-time applications are not the same as data applications in terms of tuning and ongoing management. The parameters in terms of quality and focus of troubleshooting change significantly.
VoIP phone systems, for example, are an extremely mission-critical application that must perform flawlessly. Yet the reality is that most current network infrastructures are optimized for data, not for converged data, video and voice traffic.
While data networks that achieve 99.9% availability are considered to be operating extremely well, that number falls well below what people expect from their phone systems, which generally deliver five 9s — 99.999% availability.
Another issue to consider is packet loss and out-of-sequence packets — the cause of 50% of the quality of service (QoS) problems with VoIP systems. These are issues that network managers traditionally have not had to worry about for data transfer. Packet loss has always been acceptable in data transfer, and TCP/IP is designed to handle it.
But VoIP does not use TCP, so a dropped word translates into dead air, resulting in a garbled phone conversation.
Therefore, in planning a move to an integrated IP approach, network managers will want to set the bar for acceptable quality much higher than they have been used to in the past.
This means addressing the relevant issues using a lifecycle approach that encompasses everything from network planning and initial deployment to expansion and ongoing maintenance.
It can’t be stressed enough that the challenges of IP don’t stop with the completion of the initial deployment.
A successful rollout should never lead to complacence. Organizations often find that despite a great start, they begin to experience problems when they try expanding the system to a new set of users.
VoIP conversations for example will become jittery and choppy; calls get cut off; the whole system slows down and the overall quality of network performance and user productivity suffers.
IT managers often try to solve these problems by the tried-and-true solution of adding more bandwidth.
However, when one considers VoIP performance specifically, the inherent problems generally are not related to bandwidth or throughput.
The problem is likely to be latency, which is typically responsible for the other 50% of QoS issues experienced with VoIP systems (that is, the problems that aren’t due to packet loss or out-of-sequence packets).
Even if an IP system works fine through a few initial rollouts, there is no guarantee that later network changes will not affect call quality or have an impact on other mission-critical applications.
Managing the complete VoIP lifecycle is an approach that starts with pre-deployment qualification of a network through to deployment ongoing monitoring and management, troubleshooting and planning for future growth.
The barrage of tools required for this undertaking include communications testers; protocol analysis and monitoring devices for troubleshooting problems at the network’s edge and isolating degradation and slowdown problems; hardware analyzers to gain visibility into network traffic and gain real-time packet capture and analysis; integrated network analyzers for capturing and analyzing voice and data traffic to assess QoS; WAN analyzers; and reporting devices for analyzing of bandwidth usage.
Together, these tools provide everything a network manager needs to perform the full range of IP lifecycle management functions.
Let’s now look at a breakdown of the different stages in the IP lifecycle and related testing requirements:
Pre-deployment: It is essential to put the work into ensuring a network is ready to support the demands of fully-integrated IP functionality before you lay the first cables. Issues to address include assessing its ability to meet speed and bandwidth requirements and the capability to deliver Power over Ethernet. Tools can be used to verify if all drops are active and that power is available from switches, whether there is proper communication with network services, and if all switch ports are properly configured.
This is also a good time to establish performance baselines to allow verification of the impact of additions to the network in future.
Deployment: Here a multitude of tools can be used to verify performance, monitor transmissions, check bootup and initialization of VoIP systems, setup time, and quality of service, among other parameters that are critical to a successful rollout. Again, a performance baseline measurement should be performed on completion of deployment to ensure QoS metrics are being met. This provides a reference point if and when future problems occur.
Ongoing monitoring and management: The testing job does not end with the rollout. The need to monitor such key issues as bandwidth usage and QoS is equally critical after initial deployment.
This includes taking regular performance measurements to ensure everything remains up to QoS standards. In addition, it is essential to have a means to capture and decode packets in order to analyze the origin of a problem and resolve it quickly.
Troubleshooting: Beyond the ongoing day-to-day maintenance requirements, network managers must have the tools and processes in place to affect a quick resolution to any problems that do arise. When troubleshooting, a good place to start is the network’s edge to look at communication details between the network and the server. In many cases, this can be performed at the user’s desktop.
A quick and accurate assessment at the right place on the network can lead to a quick resolution and avoid the headaches of having to escalate problems to the next level. If the network edge is not the problem, troubleshooting needs to go a step further to include real-time QoS analysis of network traffic and links, isolation of network degradation points and identification of configuration errors. With many of today’s IP-optimized test tools, much of this can be achieved without having to go through the complexities and expense of extensive packet analysis.
Planning for future growth: This is where baseline measurements prove invaluable. Studying trends and baseline data from the cu
rrent IP system helps to determine the available capacity for growth and new resource requirements. Test tools can help determine how much traffic volumes have grown over time, and what upgrades are need to support increased data flow.
Combining voice, video and data onto a single IP network also adds a new level of complexity that current monitoring tools may not be equipped to handle.
This is why, as organizations forge ahead with their IP rollouts, IT managers increasingly say they are becoming overwhelmed with the growing challenges of managing these networks, where potential problems can increase exponentially with each addition.
For example, they may notice degraded performance from specific locations at certain times of the day, but they may not have the right tools in place to drill down and figure out why the problem is occurring.
What IT managers need are “IP-friendly” management tools that provide a clear view into the network at all locations to ensure predictable, consistent performance for all applications traveling across the LAN and WAN.
They need tools that detail which applications use how much bandwidth and when, provide root-cause analyses for network slowdowns and outages, measure QoS performance in the routers and carrier networks, determine how policy affects QoS, and analyze network performance both in the main and branch offices.
In the emerging world of converged IP, it is critical that IT managers undertake a comprehensive approach to the management of the complete IP lifecycle, from pre-deployment qualification of the network through deployment, ongoing monitoring and management, troubleshooting, and planning for future growth.
Having the right tools is a good place to start.
Brad Masterson is product manager for Fluke Networks Canada and can be reached at firstname.lastname@example.org