Why we need to separate voice vlan and data vlan

[Yenosh]

Why we need to separate voice vlan and data vlan?

[Kasun Bandara]

Hi,

this is very big explanation if you need to understand A-Z. :)
i ll give quick answer,
voice traffic is a very sensitive for real time packet transmission. if any loop or broadcasting happens within network, that will affect to voice traffic. so its better isolate voice traffic from data VLAN to avoid data broadcasts or other traffic congestion from data network.

[Dennis Mink]

Theoretically you dont have to. Qos can class on rtp and sip without issues. But its a best practise to minimise layer 3/broadcast 'interferance'. As kasun mentioned.

[balaji.bandi]

This is going to long disucssion and debate : 

But personally i make some comments why you need do it (why not again design consideration).

1. Since Voice is very sensitive application ( as mentioned other post) and you want have Call Quality and Performance

Poor VoIP performance can easily hinder a business that relies on its voice communications.  A well-segmented VLAN helps ensure your voice quality is as good as possible.

2. Security, so that you can identify what is the end device and you can apply security policies based on the end point.

3. Easy to problem resolution ( so you have data and voice in different VLAN).

LAN segmentation brings significant performance and security enhancements, while taking little or nothing away from overall network flow.  

Other side, if you have Device (like PC have voice and data to be done on same device, then we need to consider other design guide lines)

[Dennis Mink]

@balaji.bandi  , like you stated a long discussion, but i would argue the case that your point 1 , is not really valid. traditionally you would QoS mark based on simple ACLs, now you can do heaps more, trust on DSCP (which you would do on an access port level, and on top of that you could mark voice/video matched on RTP. so braking VLANs up, in my opninion is not a compelling case. dont get me wrong I am still an advocate for splitting voice and data on layer 3, but not necissarily for QoS purposes

[balaji.bandi]

@Dennis Mink  yes i do step back now in regards to point 1, and realized yes, agree with you, you can mark the packets immaterial where they come from.

[Dennis Mink]

Np mate i wasnt trying yo be right. I justed want meaningful tech discussions. Always good to learn

[Yenosh]

Thank you all for helping me to understand the concept