10/half in; 100/full out?

[Esmerelda]

From Catalyst 10/Full connection to a Netgear 10/100 LAN non-programmable switch. Can I get 100/Full connections out of the Netgear? The Netgear isn't programmable, but the equipment on the "other end of the ethernet cable" can be set for 100/Full.

..........Or, do I have to change the speed on the Catalyst link?

 

[bcastner]

If autosensing between the devices has no particular issues, you should be fine.

 

[Esmerelda]

I'm sorry. The requirement for the "end" equipment is to be hard coded. I guess my question stands as to whether I need to upgrade the speed on the Catalyst link to 100/full in order to assure that I can supply 100/full to several IP elements hardcoded on the equipment end.

 

[jimbopalmer]

I have a FAQ in the ethernet forum, but basically you should never hard code either end of a wire as Full duplex, unless you can hard code both ends of the wire as Full.

If leaving it on Auto-negoitiate is not a choice, then hard code Half Duplex.

I tried to remain child-like, all I acheived was childish.

 

[Esmerelda]

Well, I guess I get confused because I'm going through a Netgear switch that is fed by a 10M/half uplink.

I'm thinking that if I need to satisfy the hard-code requirements of the end points, I should dedicate a Netgear switch (my Netgear switches aren't programmable) to a hard-coded 100M/full Catalyst switch port, and then I will be able to hard-code all of my end points that are fed off that particular Netgear switch?

 

[jimbopalmer]

Any wire you can hard code at both ends will be fine, near as I can tell, no wire attached to the Netgear can be configured at the Netgear end, so none of them can work hardcoded to Full duplex. Equipment directly hooked to the Catalyst may be configured if the Catalist is managed AND you can manage the other end.

I tried to remain child-like, all I acheived was childish.

 

[dk87]

Opinion: Throw the Netgear switches away. They are poor performers and will never give you the performance of a Catalyst.

 

[Esmerelda]

About throwing the Netgear switches away........even if they WERE good performers (as you say they are not), I think that what jimbopalmer is telling me is that I have a bad design when I place a non-managed switch in the mix with several IP end-point that require hard-coding.

 

[jimbopalmer]

Correct, you will need a switch with some way to set duplex, (DIP switches, a web page SNMP, someway.

What kind of end devices require hard coded duplex? I have not seen any sice 1996.

I tried to remain child-like, all I acheived was childish.

 

[Esmerelda]

Avaya PBX CLAN and Medpro cards. CLAN can be set to auto, but it is not recommended.

 

[jimbopalmer]

I could be wrong, but won't those also need QOS capible networks? You won't get voice quality VoIP on non QOS capible gear anyway.

(Much of my Nortel VoIP gear only did 10/Half, but it needed QOS and VLANs, I used 802.1p, as it was simple enough I understood it.)

I tried to remain child-like, all I acheived was childish.

 

[Esmerelda]

You are right about the QOS and VLAN. True VOIP is only being used on one of our sites, and is not of issue here. We are using IP Agent, CMS (Call Management Links), Call Detail Record and Voice Announcement as the majority of our services. We are driven by Avaya requirements, which dictates hard-coding. I think that we have created a monster by putting a Netgear non-managable switch in the mix, because we can't hard code both sides of the wire.

Please let me know if I am wrong.

 

[jimbopalmer]

Yes, both the fact that it will not do QOS, and the fact that you can't hard code Full duplex on it's ports will make the Netgear a poor choice near any VoIP install.

I tried to remain child-like, all I acheived was childish.

 

[Esmerelda]

Jimbo;

I feel that I need to finish your sentence to make it fit my use, since VoIP isn't even important here.....
"......or application in which hard-coding is a requirement". I'm beating this up a bit, because I'm working up the courage to approach my management with a proposal to remove the Netgear switches and home run to a "real" switch (which, in many cases there are no ports available, adds complication, and the necessity to work with the customer in upgrading their switch). Our PBX and associated IP-centric elements (Voice Announcement, Call Detail Record, Call Managemeng System, etc.) are physically located at the customer's site. As challenging as it is for "voice" people to work with their own LAN group, imagine working with your customers' LAN groups. This gets further complicated when the "voice" management can't spell "IP". I need to be convincing.

Thank you so much for helping me to understand what I have been suspicious about for so long.

 

[plshlpme]

what is your setup?

catalyst -> netgear -> client machines ????
or are you replacing the catalyst with the netgear?

 

[Esmerelda]

Catalyst 10MB link -> Netgear -> IP end points, some with the 100MB hard-coded requirement.

 

[plshlpme]

every port will autosense on the netgear.. so you can have a 100mb connection between the netgear and some endpoints your only gonig to get 10mb between the cat and the netgear though...
if your only requirement is 100full between the netgear and hard coded eqipment that will be fine...


cat ==10mb== netgear ====100mb==== "hard coded equip"

 

[Esmerelda]

OK. So, if I upgrade my Catalyst port to 100full, I won't have to remove the Netgear? And, if I had one or two end points that were hard-coded to 10half on both sides, that would work, too (without removing the Netgear?)

 

[plshlpme]

ya each port operates independently...
as long as each port negotiates properly you will be fine.
there is no need to remove the netgear.

 

[Esmerelda]

....as long as I upgrade my Catalys port to 100full?