What is the maximum size the RF can be before it becomes inefficient.
We have :
1000SM (internet, VLANS)
9 AP clusters
1 BH cluster
6 Remote AP’s
At the moment this is one big RF cloud, what size RF clouds are you guys running ? are you keeping it switched or putting routers at each cluster.
I’m confused. Are you talking about RF or IP?
Aaron
RF…
At the moment all our RF acts like a big layer 2 switch cloud. Surely as we add more AP’s and SM the bridge tables in the BH and AP must start to grow to a point where the thing starts falling apart…
Well, that really has nothing to do with the RF side of things - you are talking about a large layer 2 network. If we are talking about RF we are talking about the signal levels themselves, not the data transported.
In terms of keeping the layer 2 small, VLANs are great for that. I see that you already use VLANs for other network isolation. In my network I use VLANs to keep each tower network separate - each of my towers serve customer that are on different VLANs, one for each tower. All of my network management gear (ie. APs, CMMs, switches, UPSs) are on another VLAN. And then, of course, I have my private VLANs for customer LAN extension.
Aaron
sorry thats what I meant… the layer 2 switch how big can we grow it…
At the moment we use vlan for private customers who want to create WAN for multiple sites.
so let me get this right…
Cluster 1 AP1 = VLAN 101
Cluster 1 AP2 = VLAN 102
Cluster 1 AP3 = VLAN 103
Cluster 2 AP1 = VLAN 201
Cluster 2 AP2 = VLAN 202
Cluster 2 AP3 = VLAN 203
we don’t have switches at the towers yet… ups yes…
how does that help the canopy layer 2…
arp ?
bridge table ?
processing ?
You kinda have it right for what I do. All of my APs are on one VLAN - 15 sites. It’s the SMs that are on the different VLANs. My router is VLAN aware and I bring most of the VLANs back to it (also different IP subnets).
Doing this keeps the broadcast domains small - if there is any kind of broadcast storm or any other data layer problem it will be isolated to that particular VLAN. I have seen multicast storms from a customers location bring most traffic to a slow crawl, but again it only affected that VLAN.
Aaron
i’m thinking of putting a cisco 3600 at each location… need STP and also looking at MPLS stuff…
Not sure what MPLS is (I know, Google it. haha). STP is nice. I am using it all over my network, redundent links are great! If one fails for any reason all traffic goes the other way. It also helps when doing maintenance on the system - you don;t have to worry about rerouting everything before you work on something. That said, I’m not an expert on STP, so either I don’t know what I am doing, or there’s something wrong with my switches - it can take a few minutes sometimes for everything to get rerouted. Hmmm…
Aaron
60-120 sec convergence is what we get… but like you said we can only control the links from the main NOC’s, I want to be able to put in extra loops for load balancing inside the RF without them hitting the switches…
MPLS (multi packet layer switching) in the next big thing for switching rather then routing… it allows you more control on providing an end to end path…
not widely used/offered yet