Using 1252's on 10/100 ports instead of gig-e

[CFayNTAdmin83]

Hi Everyone. I had a quick question regarding using a 1252 Access Point with a power injector but only on a 10/100 port. I saw that Cisco and Intel have a nice white paper that shows actual throughput benchmarks with 802.11n clients in different spots. However, I didn't see any actual throughput values for the 1252 on a 10/100 port. I understand that a gig-e port is desired over fast ethernet, but like many other companies, we just don't have gigabit ports everywhere. I'm planning on installing most of the 1252's I get on gig-e ports, but a few of them are in areas where we only have 10/100 ports on the switch. Has anyone done any kind of benchmarking with the 1252 on a 10/100 port? I'd be interested to see how it scales when compared to the benchmarks of the AP on a gig port. By the way all my switches have gigabit uplinks, so only the 10/100 port would be the restriction...

[Leo Laohoo]

Hi Craig,

What will you intend to use the AP for? I mean, if it's just limited wireless then I guess 10/100 is OK. But if you plan to use the AP for high-volume traffic then it would be recommended to put more AP's around to minimize the natural bottleneck coming from the switching infrastructure.

Hope this helps.

[phillip.vansickle]

You could use a GLC-T, Copper Gigabit Ethernet SFP module...or if you have older switches with a spare GBIC slot, a copper GBIC.

Gig switches are pricey.

[CFayNTAdmin83]

The AP's that will be on 10/100 ports are in light to moderate traffic areas. The major areas will have the gig-e ports. I thought of possibly using GBIC's / SFP's to get the last three on gig ports, but I wasn't sure if there were any compatibility issues associated with that. I had some issues with using GLC-T's and the fiber interface on a 4402 WLC. The GLC-T's didn't let me connect to the WLC through port 1. I had to configure an address on the service port and use that interface for initial programming. It's not really a big deal since I'm using the GLC-SX-MM's in production, but if I run into the same problem but with the 1252's then the AP will not connect to the controller. I have two GLC-T's laying around that I can test with once I get the new AP's. I'm going to try and mimic the Cisco/Intel whitepaper tests regarding 1252 / 802.11n performance. The only difference will show my findings with the AP's on 10/100 ports. I'm sure I'll be able to tax out the AP, I just want to know exactly what it would take to make the wireless users attached to an AP like that slow down / have performance issues. I was thinking of using the mixed mode tests where I start with 1N and say 8 or 9 ABG clients. Then slowly increment the N users and decrease the legacy clients. This should tell me what mix of clients can get by using a 1252 with a 10/100 port instead of a gig-e port. I keep hearing great things about N, and the information (if accurate) shows that the benefits are quite nice, even compared to fast ethernet. However I just don't want to convert an already working wireless network to a system that may cause users denial of service / performance problems. Of course that's the beauty of testing right?