Speed test on Gigabit Gigapower varies alot

The main reason you get wildly different speed tests like that is that all of those speed tests test the entire route between the remote server and your PC. That includes the remote server, the remote server's internal network, that server's ISP, the peering connections between that server's ISP and the network between it and your ISP (if any), the peering connection into AT&T, AT&T's network, your line, your gateway, your internal network, and your PC, when the part you wanted to test is only the part in italics. Unfortunately, there's no way around that.

All speed tests work by transmitting a known amount of data and timing how long it takes to get from the source all the way through everything to the target.  The amount of data divided by the time is the rate.  Speed tests don't report speeds faster than what it enountered on the whole route, but the weakest link in the whole chain slows down the report.  It means that the best speed test you get out of many is likely the closest one to being accurate (especially if it isn't your ISP's speed test).

@DixitS  Just for your info, the Uverse RG is not a router or modem, but a Residential Gateway to provide triple play Uvoice, IPTV and Internet (not necessarily wireless).

We recommend a 3rd party router behind the RG for those users that need features of a true router.

Was it setup like this Computer Joe correct example for 2wire/Pace RGs:

This is what I did to use an "internal" router. I set my "internal" router to use DHCP for the WAN address, plugged it's WAN port in to the RG, reboot the "internal" router, let the RG assign a local address to the "internal" router and then set that address to the DMZ in the RG's management interface. When I go to the "internal" router's management interface it shows as having the same WAN, gateway, and DNS addresses that the RG uses.

I set the "internal" router to assign addresses to "my" side of the network in a different IP range than what the RG uses (192.168.2.* instead of 192.168.1.*) but using the same subnet mask (255.255.255.0). My internet works fine with no interuptions and local network tasks (back ups, streaming, etc.) work as expected.

I can also still access the RG from "my" side of the network when I need to without having to change any network settings or swap any cables. Leave DHCP running on the RG. You do not need to disable the firewall in the RG as the DMZ will open a pinhole through it to the address you pick (your internal router). If you have existing wireless on your router that your satisfied with and want to keep, just make sure to turn off the wireless in the RG.


As for the STBs they should be run straight out of the RG with CAT5 or RG6 Coax.

😉

Chris
__________________________________________________________

Please NO SD stretch-o-vision or 480 SD HD Channels
Need Help? PM ATT Uverse Care (all service problems)
ATT Customer Care(billing and all other problems)
Your Results May Vary, In My Humble Opinion
I Call It Like I See It, Simply a U-verse user, nothing more

@mibrnsurg

The Asus AC68U was setup as a DMZplus configuration.  I had it originally setup as Computer Joe mentions, but then you take that IP that the AC68U got (wihch was just an internal RG based address like 192.168.1.x) and assign it as the DMZplus.  Once you do that and then reboot the AC68U again, it basically gives the AC68U the public IP on its WAN interface.  Where first it got the 192.168.1.x address.  My STB's are hooked to the Coax as mentioned.

I didnt fully do with Computer Joe mentions because it sounds like that is basically going to give you a double NAT situation which isnt ideal when you also trying to port fwd for certain like my Security cameras, and other items.

Dixit

After re-reading again a few times what Computer Joe mentions, sounds like exactly what I did.  He said I put the "internal" router on the DMZ.  Im assuming he basically means DMZplus.  That is basically what I did.  It hands the public IP off to the "internal" or "3rd party" router at that point.

And like Computer Joe, I have different DHCP assignments, I have my Asus giving out 192.168.0.x and the ATT GW gives out 192.168.1.x if you connect to one of its 4 LAN ports (which nothing is connected except the Asus Router).

Dixit

---

@DixitS wrote:

After re-reading again a few times what Computer Joe mentions, sounds like exactly what I did.  He said I put the "internal" router on the DMZ.  Im assuming he basically means DMZplus.  That is basically what I did.  It hands the public IP off to the "internal" or "3rd party" router at that point.

 

---

That's what I thought when I read your post, but I decided I may be missing something and let it go.

So I figured this out.  Tried another machine hooked into the same Asus router which is running point since its on the DMZplus.  That machine was easily getting in the 700s both ways on the ATT Speedtest.  So tried the original server again, same issue, still barely 70 down as the speedtest shows.  So at this point knew it was down to the server.  However remember I mentioned that server pointed to Comcast or Charter speedtest easily gets much more download so its not the link in any way.

Tried to update the NIC firmware, got no change.  BTW its a Windows Server 2012 box.  So then ran across a cmd line feature for the NIC for windows and turns out it fixed it.  It all had to do with ECN Capability.

Here is a link that shows how to enable/disable ECN

http://social.technet.microsoft.com/wiki/contents/articles/20204.how-to-enable-and-disable-explicit-congestion-notification-in-windows.aspx

Basically first ran the command of:

netsh interface tcp show global

This shows all the hidden features of the NICs for Windows.  Shows whats enabled/disabled/auto

Then ran the command to disable ECN:

netsh int tcp set global ecncapability=disabled

Once I disabled it, I ended up on that same original problematic machine getting 800/950 on the speedtest (down/up) using the original problematic ATT speedtest server.  So all is good now.

ECN is normally used for large networks/businesses.  Not much use in my opinion for a home network.

According to the article you linked

---

@Microsoft wrote:

Windows versions since Windows Server 2008 and Windows Vista support ECN for TCP, but it is disabled by default

---

had you turned it on yourself for some reason, or is Microsoft wrong about it being turned off by default?

That article references Windows Server 2008 where it wasnt enabled by default.  However for Server 2012 it is enabled by default.  But the commands are still the same as far as enabling or disabling it.

Dixit

The article says "versions since..." which includes WIndows 2012 (and all future versions).  It is therefore incorrect. 

Yea its seems incorrect because I tested this by running the same command on my Windows 2012 R2 box which was recently installed, no other programs been installed on it and it shows ECN is enabled.

Dixit