Unsolved
This post is more than 5 years old
4 Posts
0
3716
Network Topology Suggestion
Hi,
i have been assigned to design a network topology with the following requirements.
Site A (main site)
80 PCs, 30 IP HD cameras, 40 IP phones, IP PBX for telephony, NVR for Surveillance
A fiber cable connect the sites together.
Site B (branch)
40 PCs, 20 IP HD cameras, 20 IP phones
Site C (branch)
40 PCs, 30 IP HD cameras, 20 IP phones
Site D (branch)
40 PCs, 30 IP HD cameras, 20 IP phones
The topology scheme below is my proposal.
Any corrections, suggestions, improvements are welcome!
Thank You.
dell-richard g
605 Posts
0
February 3rd, 2018 16:00
Although I am not into network design, maybe I can provide some general suggestions:
1. Based on your layout, the single N4032 ties all your sites together. If that single N4032 fails, then all communications between the sites will fail. Consider adding redundancy (stacked unit).
2. Is there enough bandwidth on the 2 x 10Gbps port-channel (LACP) between the switches. Maybe a third link in the port-channel. With the NVR, IP phones, and high-definition cameras, possible that the links can overloaded with just two links in the port-channel.
3. Does the single N4032 have enough system resources to handle all those IP cameras. CPU utilization..etc
4. Are you considering implementing data backup on this design?
Hope this helps
BTW: Took me a few minutes to decipher in your diagram what "stucked" meant. Stacked :)
stampk
4 Posts
0
February 4th, 2018 06:00
I am not an expert into network design too Richard, plus I have no experience with that size of networks. So, any help would be highly appreciated.
1. Backbone redundancy is very good idea. So, i have to change the design a little bit.
2. After a quick calculation, average bandwidth per pc/printer=~10Mbps, Camera=~3Mbps, phone=~100Kbps
Maximum total instant bandwidth for/from
Site A. 800Mbps+90Mbps+4Mbps= 894Mbps
Site B. 400Mbps+60Mbps+2Mbps= 462Mbps
Site C. 400Mbps+90Mbps+2Mbps= 492Mbps
Site D. 400Mbps+90Mbps+2Mbps= 492Mbps
I think that 10Gbe links can easily manage this bandwidth.
3. I used to work with N3000/N4000 only for clustering data center servers. I have no idea if this model is the right choice for this project.
4. No need for data backup.
murmanov
59 Posts
0
February 7th, 2018 02:00
Hello, your design looks good for small enterprise, but I'd like to add some thoughts as well:
1. Keep in mind once you started to use stack you're losing ability to upgrade switches without interrupting your service. That's why I suggest to use VPC ( N4032F can do MLAG without stack ) in that case you'll have separate control and data planes for core switches
2. If you don't need shared Layer2 between 4 segments, I'd suggest you to use Layer3 interconnection and built some sort of ECMP to core devices
3. Ensure that edge router sending default gateway only to core N4032F switches. In that case you'll get real core / edge / access segmentation
stampk
4 Posts
0
February 10th, 2018 03:00
Thank you very much for your helpful suggestions.
1. You are right. I used to work with stack, but mlag is better this time. And as the rule says "lag vertical, mlag horizontal".
2. Layer2 is a must between segments.
3. I am sorry but i cant understand your advise. You mean i should use Core switches act like gateway(routing) for network segments and let the edge router only for gateway?
murmanov
59 Posts
0
February 15th, 2018 02:00
Sorry for late response, I didn't get mail notification somehow.
Regarding item 3. I meant that you should keep L3 connection between core switches and edge router and send only default route information from your edge device.
All vlans should be terminated on core switches and it should act as default gateway for each existing vlans
You should configure both links to edge router as layer and it depends on you if you want to utilize both of them