AMK Networking

Thanks for visiting the AMK blog. This is where we like to talk about best practices in the I.T. industry. We are an I.T. Company based in Ottawa, Ontario, Canada. AMK designs and builds next generation secure, scalable, survivable networks: providing our customers with the building blocks necessary to ensure secure access to their information.

This time we are going to discuss some steps to avoid some very serious Voice over IP(voIP) disasters. When clients engage us after their VoIP deployment ended with poor results we hear some similar complaints. “Everything was fine before we put VoIP on the network”; “We have Gigabit in the core and 100 megabits to the desktop so we know the network is fine”; “We were under pressure to deliver VoIP so we just put it in the network”; “We had the network certified for VoIP 2 years ago so we thought it was fine for VoIP now”. This is almost always followed with, “It must be the VoIP equipment because everything was fine before we put it in the network!!” In every instance the root cause was not the VoIP equipment, rather it was the IT infrastructure. The problem was always there, it was just not visible until a real time application was put on the network. So what were the real problems? Here is a list of our top 5:

1. Spanning Tree – This is one of the least favorite protocols on a network, but it does serve a purpose. It is not so much Spanning Tree itself; it is how it is implemented. Spanning Tree ensures that there is one active path between any two devices in a network – which is great. The problem occurs when there is a topology change that affects the network infrastructure. The convergence of the network can take from 30 seconds to 2 minutes. It is worse when Spanning Tree is disabled and loops are created. Now there are multiple active paths which create broadcast storms which will consume all the bandwidth in the network. Another problem occurs when multiple spanning tree groups exist and there is a connection between groups. Again a loop is created and the network experiences various issues. One client had experienced over 100,000 topology changes in a month!! When they added VoIP on the network failed! It wasn’t Spanning Tree itself but how it was employed.
2. Quality of Service – No matter what, Quality of Service (QoS) must always be used in a VoIP network must be designed well; especially in a multi-vendor environment. This is so critical for success. Note that each vendor may apply different values and queues for Voice traffic. We had a client that had two different vendors and accepted the default values for QoS and queuing. The result was one vendor’s VoIP traffic QoS values were the other vendor’s routing traffic QoS values. The users’ experiences intermittent call drops.
3. VLAN contamination – This is a term that we coined to describe when one VLAN is directly connected to another VLAN through VLAN unaware ports. Imagine a VLAN that has a high level of broadcasts or multicast connected to a VoIP VLAN. Two things happen. One is the bandwidth is consumed and the second is that each phone or telephony device must process the broadcast or multicast to then discard the packet. The users’ experiences jitter and call drops. Iif QoS is employed now the broadcast/multicast traffic is sent through the network with a higher level of service.
4. Port mismatching – Autonegotiation is a great feature, it has the ability to identify the data rate (I do not like to use the word speed as that is a misnomer) on the receive pair and negotiate the best data rate to use; whether it is 10/100 Megabits or 1 Gigabit Ethernet. Autonegotiation cannot determine the duplex of the connection. If both ends of a connection are not configured for the same operation – hard coded or autonegotiation – the outcome is a low performance connection. Some devices require a hard coded configuration. One client connected a signaling server which was hard coded to a switch port configured for autonegotiation. The mismatch caused calls to be dropped or call setup failure. In other cases the users’ experienced jitter and significant delay.
5. Grounding – Yes Grounding. A client using a multivendor solution in which both vendors were adamant that the other was at fault. What we found was that there was a grounding issue on the core switch. There were 3 volts on the ground. The users experienced a wide variety of issues from static on the call; one end was loud while the other was quiet, to loud screeches then the call drops. I also include static shocks in this category as proper grounding and static wrist bands prevent static shocks. Why is this important? 50 percent of all static shocks result in immediate failure. The other 50 percent will fail over the next 5 years. This is the really important part. MOST manufacturers will null and void your warranty if failure is due to improper grounding or static shock – and they will find out.

So how can you avoid a VoIP disaster? Well, first of all do not make assumptions about anything on your network.  Second, if you have well documented processes then follow them diligently. If you do not, here are some suggestions from the process that we follow. As with everything we do this list is a living thing and should be reviewed and adjusted as needed.

• Develop, maintain, and follow a project plan. This is the only way to measure success of your project.
• Perform a thorough gap analysis of your IT infrastructure. It does not matter if a vendor or reseller gave it the gold stamp 2 years ago. Your network has most likely changed in that time.
• Review QoS and the architecture
• Design a VoIP infrastructure based on the results of the gap analysis.
• Build and use a lab to test both the recommendations of the gap analysis and the design.
• Make the necessary changes in the network
• Pilot and review the VoIP technology.
• Document everything. This includes test plans, implementation plans, configuration files, topology diagrams, everything!! This step trumps all other steps.

Well that is it for this time. I hope you have enjoyed our post. If you have suggestions, questions, comments, or challenges please send a note at blog AT amknetworking DOT com. If you think we can assist your company or organization, please Request A Consultation with us.

Tags: Computer networks Ottawa, Ottawa I.T., Ottawa I.T. services, Ottawa I.T. Solutions, Ottawa Information Technology, Ottawa Network Technology, VoIP Assessment, VoIP Best Practices

Posted on Friday, November 20th, 2009 at 7:21 pm and filed under Thoughts.
You can leave a response, or trackback from your own site.