Add a second router to extend network?

Hi,

I have a fairly new Netgear Rangemax 240 that does a good job sending the signal throughout half of our property.

I have a second Rangemax 240 still in the box that I planning to setup in the other half of the house to extend the wireless signal (it really drops off in some places) and also act as a wired hub for another PC in that part of the house.

I figured that, even though I don't have any devices that can exploit the extra bandwidth, the 240Mbps signal between the two routers would be better than running a 100mbps cable from one room to another.

This setup seemed like a good idea at the time, and the salesman told me it would work, but now I'm not sure how to go about setting up the second router to get this to work.

Any suggestions before I start, or does this sound stupid?

Thanks.

Reply to
Pupkin
Loading thread data ...

Pupkin hath wroth:

formatting link

The 240 is the speed of the wireless link. You won't get a file transfer rate of 240Mbits/sec. Various tests show about 100Mbits/sec maximum.

formatting link
the thruput tables and graphs.

You want to use the 2nd Rangemax 240 as an access point, not a router. The Rangemax 240 apparently does NOT support WDS (wireless distribution service). So, you must a CAT5 cable from the main router to the added access point. See: |

formatting link
setup instructions.

Reply to
Jeff Liebermann

No. Cat5 cable is better

The wireless speed rating is a "best of all possible worlds" number - the cable 100M is actually achievable - and the cable is 100M full duplex, whereas the radio is half - ie only 1 device can send at 1 time.

in reality 802.11 and all its variations has a lot of overhead to manage the radio specific bits - this normally limits you to 50% useful traffic across the available bandwidth - and in many situations the speed rate adapts to a lower setting as well.

Also for the 2nd device to act as a repeater, each repeated packet has to cross the wireless channel twice - so you at least halve the thruput on top of that.

Cables are usually better where you have a useable route - useful for other things well, more reliable, tolerates interference and so on better, faster. And always run more cables in than you think you need.

Reply to
stephen

Major nit - While the chances of getting a full bandwidth connection are certainly better with wire, there is no way in hel? that any 100 Megabit wired connection will reach a full 100 Megabit. For the same reason you mention

You forget about the 8 byte sync of Ethernet packet that precedes the MAC address, and the required "Inter-Packet Gap" required to detect the end of one packet, and the beginning of the next. Minor, but it adds up. Likewise, not that many network stacks can tolerate a continuous stream of "end-to-end" packets. 90 Megabit is a more realistic absolute maximum transfer rate using a cross-over cable, and the sustained rate is going to be dependent on the computer hardware at both ends, and what they are doing. Sure, a Pentium with a 64 bit wide PCI bus is _capable_ of a Gigabit/second, but only if it's not running an O/S, or user applications. Most systems aren't used that way (32 bit color icons and other eye-candy take a tremendous amount of CPU cycles - look up "Metcalfe's Law"). If you are not using a direct cable, then you have to take into account the capabilities of and load on the Ethernet switch or other active device located between the two end systems.

Old guy

Reply to
Moe Trin

Agreed.

But the difference for a cabled connection is minor.

i worked on a newspaper system (a long time ago). The interfaces between the image processor and raster engine ran for several hours a day at 99.4% load.

Since this was with nearly all full size Enet frames and the layer 1 overhead of inter packet gap and preamble of 20 bytes on a full size frame amounts to 1.3% or so - so the layer 2 useful load was getting up to 98%.

given the overheads in wireless where running at 50% useful bandwidth is considered best case, and using a device as a repeater would halve that - 2% overhead disappears into the noise....

yes - and one of the big issues when you try to work at the limit of any packet based system is exactly what does "wire speed" mean.

for example the clocking on an Ethernet port is +/- 100 parts / million - so a "wire speed" switch may run 100% in and out, but still drop packets (long on going argument with a customer).

Agreed again.

But many switches are as near to wire speed as makes no practical difference (i recently ran some tests on a box for work with up to 20 * GigE - it ran

99% load full duplex on up to 18 at a time).

Since many "server optimised boxes" can fill a GigE pipe (or 2) - then 100M is achievable.

more to the point networks may carry the aggregate of streams from many devices - so it is good practice to make sure they dont impose extra bottlenecks unless there is a good reason or tradeoff involved.

whether this matters for a home system is a different issue.

Reply to
stephen

Cabling-Design.com Forums website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.