cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
648
Views
0
Helpful
9
Replies

network design

Hello everyone,

 

I have designed a network topology for my company, and my network will carry all type of traffic Data, voice and Video (IP camera). you will find an attached file that shows my topology. my topology has been designed in hierarchical approach access switches and distribution layer three switches. as you can see from my topology, i have dedicated one layer three switch for Data and voice and the other for Video. my point is, if what i have done is good practise. if it's not can any one suggest best solution.

Thanks  

2 Accepted Solutions

Accepted Solutions

Mark Malone
VIP Alumni
VIP Alumni

Hi

it looks like a collapsed core topology which is fine too, are you going to use a fhrp or you could end up with single point of failure ,the choice is yours but I would have the vlans for voice video data on each core switch , so if one fails it can still break out of the network through the other switch , if video switch fails all video traffic fails ,if you use hsrp you can still have traffic go through one switch as active just for video subnet and send voice data through other switch as active gateway but at least if one switch fails traffic can be resilient and move across to the other switch or maybe you already are but its just not shown in the doc

View solution in original post

Hi

if you put qos and multicast in place for your video traffic it wont be a problem , you may not even need the multiast we got our vendor to use unicast video instead cut down the volume of traffic from HD cameras  , I have 2960s around my campus buildings with hi def security cameras on them each  , all they run is standard qos auto qos trust dscp on access and uplinks that's enough to prioritise the video traffic coming from the cameras , I don't even have multicast set and they work fine clarity is perfect , I do have more powerful core switches but we have hundreds of users , our NVRs are connected to couple of 3650s in 2 different security rooms too no issues

you only have 10 cameras set per switch at access I have 20 on one of 2960 switches bit overkill with dual uplink in fibre but they work fine

you could always add a couple of more pipes to the uplinks to the core switches , bundle them to together for port-channels for more throughput and extra resiliency

View solution in original post

9 Replies 9

Mark Malone
VIP Alumni
VIP Alumni

Hi

it looks like a collapsed core topology which is fine too, are you going to use a fhrp or you could end up with single point of failure ,the choice is yours but I would have the vlans for voice video data on each core switch , so if one fails it can still break out of the network through the other switch , if video switch fails all video traffic fails ,if you use hsrp you can still have traffic go through one switch as active just for video subnet and send voice data through other switch as active gateway but at least if one switch fails traffic can be resilient and move across to the other switch or maybe you already are but its just not shown in the doc

Hi,
What you have said is right and i think this is what is recommended, however, the reason why i have done it like that, i want to  separate physically my video traffic from other traffic because i am afraid that it will slow down my network. as you can see i have a very considerable amount of video traffic and on each zone i will create video VLAN. if you have any good suggestion about how to handle the video traffic fairly.
you help is greatly appropriated.

You need to look at QOS to prioritise the video traffic within the LAN.

Your distribution switches should really act as a pair ie. they should both be responsible for all vlans otherwise as already pointed out you have a single point of failure. In addition if you want to do maintenance, upgrade IOS versions etc. it would allow you to take down one switch while the network stays up.

Jon

personally first I would scope out how much exact video traffic your pushing versus what the switches can actually do per the data sheets make sure your not overloading them , test it with the redundant design see if your video traffic actually effects the normal traffic , if it is an issue you could just use qos at layer 2 , AF41 is video traffic and EF is voice traffic , if you just set either mls qos triust dscp or auto qos trust dscp on the acces switches it will segregate the video and voice traffic from each other in queuing systems , EF will get the highest priority then video next , that will ensure each is prioritised and you can tweak it if you have to , to allow or remove more bandwidth

what type of access switches are these and what are the cores the exact platforms

Hello,

just to be exact and clear, that network design has not implemented yet and it is for industry zone as well as all the links between the switches are going to be in fiber optic due to the long distances. i am planning to use cisco 3850 series in the core and the 2960 in the access layer. it going to be about 100 IP camera throughout all the factory and all that video traffic is collected in the NVR (network video recorder) located in the zone Z04.03 as shown in the diagram. so, there is multiple users want to see IP cameras from different locations which i think is going to increase the amount of video traffic. i am not sure if my network could resist all that.

your help is greatly appreciated.

 

Hi

if you put qos and multicast in place for your video traffic it wont be a problem , you may not even need the multiast we got our vendor to use unicast video instead cut down the volume of traffic from HD cameras  , I have 2960s around my campus buildings with hi def security cameras on them each  , all they run is standard qos auto qos trust dscp on access and uplinks that's enough to prioritise the video traffic coming from the cameras , I don't even have multicast set and they work fine clarity is perfect , I do have more powerful core switches but we have hundreds of users , our NVRs are connected to couple of 3650s in 2 different security rooms too no issues

you only have 10 cameras set per switch at access I have 20 on one of 2960 switches bit overkill with dual uplink in fibre but they work fine

you could always add a couple of more pipes to the uplinks to the core switches , bundle them to together for port-channels for more throughput and extra resiliency

Thanks,

i will contact you for further assistance.

The 3850 series support stacking and 10g, so might the 2960s depending which generation of them you're using.  With such hardware, you shouldn't really need to build out application specific design.

Joseph W. Doherty
Hall of Fame
Hall of Fame

Like Mark and Jon, I too would try to avoid single points of failure.  Also like them, I too would suggest using QoS if you need to guarantee performance on a converged media network.

What I would also suggest you consider, if your core devices support stacking or VSS (or vPC), you use that with your edge switches.

If you avoid STP blocking, you may effectively double your bandwidth to your edge switches.

If you need even more bandwidth, consider Etherchannel or using 10g or mGig.  These all can be done over copper, which can bring your costs down for ports and transceivers.

If your edge switches support stacking, you might consider using that too.

Getting Started

Find answers to your questions by entering keywords or phrases in the Search bar above. New here? Use these resources to familiarize yourself with the community:

Innovations in Cisco Full Stack Observability - A new webinar from Cisco