This is challenge #1 in our blog series on fine-tuning your SIP infrastructure.
Before voice travelled over IP, the humble switch didn’t have much thinking to do. YouTube streaming, or opening large files across the network, etc, were challenging, but all bytes were equal, with no special treatment required.
Voice is different. In order to have an intelligible conversation, all packets must arrive in good order. Contention for bandwidth can cause packets to be dropped. This causes a stuttering effect, and can render voice unintelligible, or at worst, kill it altogether.
Despite the rise of email, chat and other media channels, voice communication is still by far the largest part of a contact center’s business. The quality of that voice communication is paramount.
Managed switching enables the call center to deliver SIP Quality of Service (QoS). QoS refers to both the reliability and the speed of packet delivery. Both are critical in the call center.
- Reliability of delivery
A network switch (managed or not) holds packets of all types in a queue (known as a jitter buffer) while it considers where and when to send them. But this space is limited, and at a particularly load, that space will be filled. What should it then do with new packets? In order to make room, it must drop from memory one of the least important packets. The advantage of a managed switch is that voice can be given highest priority. This ensures that if packets must be dropped, they don’t contain voice.
- Speed of delivery
If voice is prioritized, any voice found in the queue will be sent ahead of anything else. We will look at this in more depth in our next blog on voice delay.
The cost of managed switching
“But isn’t all this expensive?” you ask. Yes, a managed switch can be around 3 times the price of unmanaged of similar spec. Therefore, your finance director will tell you that parts of your business really don’t need a managed switch. Our advice is to tread very carefully. Under some circumstances e.g. where back office staff are totally isolated from call center staff, an unmanaged switch may do the job.
Switch capacity is also worth a mention here. There will be pressure to keep costs down, and buy the smallest capacity switches you can get away with. But, as in most fields of IT, the need for bandwidth is always increasing. So the rule is: plan for future growth, and always buy (much) more capacity than you think you need. This is especially true of backbone.
If you want to be sure that the voice services you are offering are of the highest possible standard, then high-capacity managed switching, throughout the enterprise, is the only way to go. For centers that are migrating from TDM, this may mean ‘rip and replace’ – a painful but necessary process.
In our next blog, we look at SIP challenge #2 – voice delay.