Wednesday, March 14, 2012

Home Networking - Wireless-N vs. CAT5e vs. CAT6

With Internet Service Providers cranking up the speeds available to home users, and home media servers becoming more popular, it's starting to become more important in what kind of networking you use at home.   Otherwise you may be paying for speeds you can't even achieve with your hardware.


First off, if you are still using a Wireless G router (54mbps), and have anything better than DSL, it's a pretty safe bet that your wireless device Internet speeds are being held back by your wireless router.  With that being said, prices of Wireless-N type routers have dropped to under $30, so there's really no reason not to upgrade considering you probably pay over $50 a month for your Internet.  Also, while looking over the results, remember that ISP's like Comcast & Verizon are offering internet speeds now of over 150mbps!

I decided to bench test a lower priced wireless-N router, versus hardwired networking connections via a gigabit switch with both CAT5e and CAT6 cabling.  Here is my very non-scientific setup...

Equipment:
Wireless N Router: D-Link DIR-615 w/ newest stock firmware.
Gigabit Switch: Trendnet GreenNet 8-Port
ISP: Comcast 10mbps constant, 36mbps boost
Cabling: CAT5e 350Mhz Rated & CAT6 550Mhz Rated (CCI Bulk with self crimped connectors, used shielded connectors on CAT6) Cables cut to 30 foot lengths.
Hardware: Two i7-920 Desktops with 6GB RAM & Realtek PCI-E Gigabit network controllers. Jumbo frames set to 9K.

Methods:
Used LanBench opensource network speed testing software to take hard drive speeds out of the equation as much as possible.  Ran five 30-second tests of receiving, and sending 100MB packets over 2 connections.
Out of the five tests, I took the average of the top 3 scores.
 

Results (ALL SPEEDS IN MEGABITS PER SECOND UNLESS NOTED OTHERWISE)
D-Link Wireless N Router via WIRELESS (Open view, about 10 feet away from router)
Send: 50.47
Receive: 64.16

D-Link Wireless N Router via WIRED CAT5e (100mbps ports on router)
Send: 60.86
Receive: 93.4

Trendnet Gigabit Switch WIRED via CAT5e (1000mbps full duplex ports)
Send: 886.73
Receive: 939.44

Trendnet Gigabit Switch WIRED via CAT6 (1000mbps full duplex ports)
Send: 890.42
Receive: 945.54

Conclusion
The performance of wireless was all over the place during the tests.  For comparison, I tested a Netgear WNDR3700 wireless N router and did achieve better results, but it also costs 5 times as much.  Overall I don't see wireless as a complete solution.  It's great to have for smart phones, and laptops.  But having a desktop running on a wireless connection just doesn't seem logical.  I know some have issues with location, and that's why they invented PowerLine networking (google it). As for CAT5e vs CAT6,  unless you want to future proof your home network for many many years to come, or the need to have a few 200+ foot runs of cable in your mansion, it just doesn't make sense to pay 50% more in cabling and connectors for CAT6 at this time. 

Importance of Jumbo Frames
For curiosity sake, I attempted the wired gigabit tests with Jumbo Frames set to disabled, and performance took a complete nosedive.  What just got me nearly 950 Mbps, dropped to under 270Mbps, simply by changing that one setting in the NIC card.  The important part of this being, Gigabit cards are set to have jumbo frames disabled by default since some older hardware can't handle it, and enabling them may actually hurt performance.  But it's a simple check in the manual or the website of the manufacturer to find out.

No comments:

Post a Comment

Post a Comment