Why is your E3200 twice as fast as mine?

[SVanDee]

I have a Dell Dimension 510 desktop PC and a Dell Inspiron 9300 laptop PC networked through a Linksys E3200 router. The desktop PC is connected to one of the E3200’s LAN ports with its built-in, 100 Mbps Ethernet adaptor and an Ethernet cable. The laptop PC talks to the E3200 with a Linksys AE1000 wireless adaptor. Both PCs are running Windows XP SP3 with automatic updates turned on. The E3200 firmware is v1.0.01 build 3 dated Apr 21, 2011. The AE1000 driver is v3.1.2.0 dated 5/6/2010.

My DSL WAN connection is only 1.5 Mbps so I don’t need a lot of WAN/LAN/WAN speed. However, I transfer large files between the PCs so I would like a fast PC/PC link. I’ve been playing with the AE1000/E3200 wireless link’s freq band, mode, channel, bandwidth and physical locations. for a couple of weeks and can’t get better than 25-30 Mbps transfer rates between the PCs even with the AE1000 sitting right next to the E3200. That’s only about half the WAN/LAN/WAN rates you reported in the "New To The Charts: Cisco Linksys E3200 High Performance Dual-Band N Router" review. I’m wondering if something in my basic configurations accounts for the difference:

- Would you expect the transfer rate between LAN nodes to be comparable to the WAN/LAN and LAN/WAN rates you test?

- My desktop PC’s 100 Mbps Ethernet adaptor is faster than the wireless link but could it be limiting the end-to-end transfer rate? (The transfer rate is 94 Mbps when the laptop is also 100 Mbps Ethernet’ted to one of the E3200’s LAN ports.)

- Do you have enough experience with the AE1000 to say whether it should perform as well as whatever wireless adaptor you used in your tests?

- I’m measuring transfer rates with IPERF software running on both PCs. Are you familiar enough with IPERF to comment on whether its results should be comparable to IxChariot’s?

Thanks for any insight you can provide.

 

[stevech]

25Mbps or so is the best-case for '11g in the 20MHz channel width mode. Of course, your clients need to support 11n to do more than 25Mbps or so with an ideal signal strength. The clients may show a strong signal from the access device but it is possible that one or both clients have a weak signal TO the access device.

To increase that to 40MHz, the client devices and the WiFi access device (WiFi router?) need to support 11n at 40MHz. Note too that some WiFi products elect to not use 40MHz if there are 20MHz WiFi systems in range, to not hog bandwidth.

Lots of large file transfers, intra-LAN, need wired connections! Using cat5 cables or HomePlug or MoCA as alternatives to running cat5.

 

[woodscomp]

Steve hit on a good starting point, with the wide band etc.. I would try on the 3200 to set the wireless bands manually and uncheck the auto settings within the wireless configuration. Sometimes in mixed mode on wireless on the Linksys routers will automatically put you in G mode and will disable N especially if you have any legacy devices sharing the wireless connections.

 

[SVanDee]

I’ve tried a lot of combinations of freq band, mode, bandwidth, channel, physical location and geometry over the last two weeks. But before diving into the details, I thought I’d ask what performance is possible with an E3200/AE1000 link?

I thought someone might tell me that the differences between Tim’s equipment configuration or measurement methods and mine explain why he reported about twice the transfer rates I’m measuring. For example, maybe routers give WAN/LAN/WAN traffic priority over LAN/LAN traffic so LAN/LAN transfers will never be as fast as WAN/LAN/WAN transfers. Or maybe IPERF doesn’t define/measure transfer rate the same way IxChariot does. Or maybe the AE1000 simply isn’t as fast as whatever wireless adaptor Tim used.

On the other hand, if someone could tell me they are achieving 50-75 Mbps like Tim reported doing LAN/LAN transfers with E3200/AE1000 equipment, then I'd know my system should be able to do better.

 

[stevech]

But have you configured the router so it can go to 40MHz mode despite that it will create interference to other nearby 11g or 11n/20MHz mode WiFi networks? "Mixed mode" is one such setting.

The 20/40MHz constraints apply to all makes/models.
Some routers are more "polite" than others, i.e., they decline to use 40MHz to avoid hogging bandwidth. Some routers have a setting to override the "be polite" defaults. The naming of this option varies.

If your WAN is much lower capacity than the LAN side, WAN traffic shouldn't affect the LAN performance very much.
LAN to LAN on wired is not routed; it's just the switch performance in the router. If you want better, then use an external switch.

WLAN to LAN speeds would be constrained by WiFi channel bandwidth (20/40MHz), modulation rate (the indicated air link speed) - which is mostly the received signal strength in the FROM CLIENT direction, and the goodness of the WiFi client devices.

WLAN to WLAN speeds can be affected by the router's WiFi setting to enable/disable peer to peer traffic via WiFi. This is a security setting.

 

[thiggins]

Some of this may repeat other replies.

IXChariot is very efficient at transferring files, with little overhead. On the other hand iperf/jperf can provide very different measurements, depending on how they are configured.

I suggest you FTP or copy a multi Gigabyte file to get a more realistic idea of your network performance. Also check the methods and tips in 5 Ways To Fix Slow 802.11n Speed.

I don't measure wired LAN-LAN performance because it really doesn't vary. If your computers are current, you'll get > 90 Mbps from 100 Mbps Ethernet and 500 - 800 Mbps from Gigabit.

The test results show that I was able to get mid 60 Mbps with traffic from wired client to wireless (downlink) and about 10 Mbps less for uplink. This was with default 20 MHz bandwidth and WPA2/AES connection and strong signal levels. Using WEP or WPA/TKIP will limit you to 802.11g speed.

Be sure that your client is reporting 130 Mbps link rate when it's in the same room. If it is showing 54 Mbps maximum, something in the settings on the router or client is limiting to 802.11g speed. Also keep 10 feet or so between router and client. Closer can overload the receivers and reduce speed.

 

[SVanDee]

You've given me several more things to try. Thanks. It may be a couple of days before I can report back. My wife keeps the laptop in another building from where the desktop and router are located. I'll have to wait until she's going to be gone for a few hours to borrow it.

 

[SVanDee]

All is well. Basically a measurement problem. Tim's post led me to QCheck. It reports 45-63 Mbps depending on the freq band, mode, bandwidth, etc. That's close enough to what was reported in the E3200 review that I think my E3200/AE1000 link is performing as it should. Thank you all for helping so promptly.

I see that there's a lot of information about useful tools on the SmallNetBuilder site now that I know some terms to Search for but a Forum dedicated to what tools to use when and how would be helpful, don't you think? I'll suggest it in the Suggestion Box Forum.