Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Internet ridiculously slow, anything I can do?
- Thread starter dabills80
- Start date
- Sort by reaction score
How many students?
Exactly. Tell your school to get more T3 lines into the local network (Hey, what's that one protocol above T3?).
Either that, or do all of your download work at night when no one is on.
Exactly. Tell your school to get more T3 lines into the local network (Hey, what's that one protocol above T3?).
Either that, or do all of your download work at night when no one is on.
You are likely to get more latency going off network for DNS than using the local servers.
S-
Yes, but there is something wrong with OS X. It has huge delays resolving addresses with many servers/NAT-routers, but OpenDNS doesn't seem to experience the problem.
Uh... says who?
I've used OS X for many, many years with a variety of NATed (and NAT-free) setups and I have yet to experience this problem.
Both of you should search the forums on slow Internet.
10.5 started doing some kind of DNS bug test query that 10.4 does do. (10.4 didn't do it repeatedly because it had looser standards for valid responses.) 10.5 does it repeatedly. My guess is that some home routers run out of connection slots for the DNS queries, and some ISP/DNS servers are doing an anti-Denial of Service rate limiting on the 10.5 traffic. When you switch to OpenDNS, 10.5 gets whatever answer it needs, and stops DoS-ing itself.
Update: Oh, I see that the inventor of Bolo works at Apple (Stuart Cheshire), and reported this particular scenario bug was fixed in 10.5.5 (by reverting to 10.4 standards). So if it is still happening to people 10.5.5+, it is happening via some other path, but with the same outcome.
Update2: This guy has a great summary on the Slow Internet problem:
http://installingcats.com/2008/06/05/slow-internet-with-leopard/
10.5 started doing some kind of DNS bug test query that 10.4 does do. (10.4 didn't do it repeatedly because it had looser standards for valid responses.) 10.5 does it repeatedly. My guess is that some home routers run out of connection slots for the DNS queries, and some ISP/DNS servers are doing an anti-Denial of Service rate limiting on the 10.5 traffic. When you switch to OpenDNS, 10.5 gets whatever answer it needs, and stops DoS-ing itself.
Update: Oh, I see that the inventor of Bolo works at Apple (Stuart Cheshire), and reported this particular scenario bug was fixed in 10.5.5 (by reverting to 10.4 standards). So if it is still happening to people 10.5.5+, it is happening via some other path, but with the same outcome.
Update2: This guy has a great summary on the Slow Internet problem:
http://installingcats.com/2008/06/05/slow-internet-with-leopard/
The "problem" is/was not widespread. In fact, I would say it is rare. The problem would not occur using the vast majority of real DNS servers. It appears to only happen when the user's router is acting as a proxy for real DNS servers. Apparently, some routers acting as DNS proxies don't support the SRV DNS query.
I have owned and operated an ISP since 1995 and I have never seen a user set the DNS address to the same address as the router. I am not aware of any ISP that tells the user to set the DNS server IP address to that of their router.
S-
I have owned and operated an ISP since 1995 and I have never seen a user set the DNS address to the same address as the router. I am not aware of any ISP that tells the user to set the DNS server IP address to that of their router.
S-
It is 100% for the people who experience it. Nobody is particularly impressed by your dismissiveness of other peoples problems. I have never experienced the problem either, but that didn't stop me from taking it seriously when it was reported on MacRumors Forums.
Some NAT routers give out their own IP as the DNS server field. Obviously not that smart, but then again, they aren't Cisco routers.
Let's not play this game. Your initial statement seemed to imply that it was a widespread problem. It's not. In fact, it's a fairly obscure bug that's triggered under a specific set of circumstances. I'm not disputing that there was a bug, but I am disputing that it was widespread. I'd be willing to bet that the vast, vast majority of Mac OS X users have not experienced any substantial problems because of this bug. If it really were a widespread problem I'd be willing to bet that it would have been patched long, long ago.
This has been a message from the Department of Redundancy Department.
You'll want to amend your argument to actually have an arguable point.