-
Posts
6 -
Joined
-
Last visited
-
Speed Test
My Results
Posts posted by Ricow
-
-
Txs!
But i tried all those tips, nothing works...
It doesn't matter to how many servers i connect to, it's not the bottleneck. The tests i've performed were based on multiple sessions to exactly the same server, which results in full bandwidth usage. Take a look:
Now one session only:
And here goes what happens when trying to watch a YouTube video. The only session carrying data is the selected one. Take a look at the sidebar's network traffic gadget to see how slow it may get:
I bet it's not youtube's fault because it works like a charm in adsl. Downloading slowly in browsers is not what i care about, the biggest issue in here is that i spend all day long in streaming websites like youtube or vimeo, so that's a big deal.
I tried using brand new installations of Vista Ultimate, XP and Mandriva. There seems to have nothing i can do to do that trick...
-
Cool ! Did you get cablenut yet ? Works really well, you can try different settings and see how it goes for you.
Sure! I did try that one, but it doesn't work...
Right now I'm trying TCPOptimizer out. Also, i updated my lan card drivers sooner, but nothing changed at all.
Btw, i opened a ticket with my ISP's Help Desk, but they are not going to take a look at that before wednesday because of carnival holidays...
-
Yeah i'm not intended to keep both of the links, i just didn't cancel my ADSL plan yet because of those speed issues.
I'm trying out the tips in that link you posted.
Txs a lot, dude!
-
Txs, mudmanc4! I appreciate yout attempt to help me out!
No, i did not connect them at the same time. My DSL link is still available, so in order to compare speeds, i just unplug the cable modem wire (RJ45 in the network adapter) and plug in the DSL modem cable in the same adapter. I did that just to check why almost everything is so slow in cable, then i figured out that it only gets to download fast when there are plenty of sessions passing data (like when using a download manager software). Browsing, watching videos or speed testing is quite sluggish all the time!
As an example, if i watch a YouTube HD video, it tops out at something like 120KB/s (which is not enough to buffer pretty much any HD video in there). However, if i pop up like 5 different HD videos, it totally use all the power this link offers (~480KB/s). I know it's not some YouTube limitation because i can get 260KB/s from a single video when using DSL.
Another example: Right now i'm downloading a random linux distro (mandriva) from a server with FDM at 480KB/s. But if close download manager and start downloading from a browser like firefox of chrome, it will not download faster than 70KB/s unless i unplug the cable link and plug in the DSL link. In this case, it doesn't matter which application i use to download the file, it's always gonna use all that this 2.5mbps DSL is able to deliver (something like 240KB/s).
Weird, huh?
-
Hy everyone!
I've just switched from my ADSL2+ 2,5mbps link to a 4mbps cable. The point is that i can't reach full throughput when very few TCP sessions are on.
When I try downloading through 20 sessions, both the DSL link or Cable manage to reach full speed (~245KB/s and ~480KB/s respectivelly). Nevertheless, when i limit the same download to only one session, DSL can still get full throttle, whilst cable won't go any further like 80KB/s... I guess that's why heavy traffic pages like YouTube HD or Vimeo got so sluggish on my cable. Even TestMy.net speed come down from 2200kbps to 200kbps... yeah, that sucks!
I performed such tests in FDM, confirmed with netstat command and of course, by pluging both modems separatelly.
Any suggestions?
Txs so much!
Are they cheating on me?
in Networking and Hardware
Posted
Hy everyone!
I'm about to sue my cable company over some failures, but i need to know your opinion about something else.
Wherever i trace route to, there's always a common route where the data passes through. Should I ping any IP in the world, it's always going to hit first 10.138.0.1 (CMTS Interface) and then 10.11.0.211 (gateway to the Last Mile as they don't own a backbone) and so on. Let's call that as "network #1".
Well, except for these 4 IPs addresses. 3 out of those 4 are the DNS's, whilst the last one is the IP to the server WHERE THEIR SPEED TEST FILE IS HOSTED! The only common hop between any route and those 4 IP addresses is the CMTS (of course). Even the gateway to the internet is not hit when pinging those 4 IPs, as it goes through a different router (10.11.0.202)! Let's call this as "network #2".
Such server is hosted in a data center 400 miles aways from here. How can it be directly reached from a local link, crossing two backbones, without hitting a single hop? It must be a VPN!!! I bet they put the test file in a private network so that they cannot be blamed for "external network" issues. That's cheating!
Proof? Lets:
The very first windows are trace routes towards those 4 IP addresses (take a look at the second hop):
The following two IPs resolve to the same domain "static.corp.wayinternet.com.br". Nevertheless, they go through Network #1:
The following addresses are in the EXACTLY same block of IPs as the ones in Network #2 (200.150.13.1-255):
I also trace routed the IP where test file is hosted from a server i own in Dallas, take a look at how many Telemar's (data carrier and DC owner) routers get reached. I know they are probably not inside the DC, but it doesn't matter where you are from, there's no way to reach the facility without getting into their backbone:
Anyone has another explanation?
Txs a lot!