Is it just my computer? Honestly, I have never seen the boards this slow. Even at 8:30 a.m. Denver time, when I usually have no problem viewing or posting. But all day today it has been a bear to even see a thread, let alone post. I’m amazed I was able to put this one up.
Anybody know if something’s wrong?
Me, too. [sigh] Mondays are usually bad, anyway.
As I said a few times, you are not connecting right to the board, you are going through a network of hubs & computers, they all have to be working right. My own path to the board shows 16 computers/hubs.
I know it looks like you are connecting right to the board, but you aren’t, see?
Speaking as someone who knows a bit about how the Internet works, and who should, in fact, actually be writing a dns client right now, I certainty that when the boards are slow, it is not really due to those 16 or whatever hops to the page, but to the speed at which the server proceses my requests and sends data to me. When the boards are slow, it’s because the server is doing something slow, generally due to heavy traffic. For example, accessing and updating the database constantly. While I’m not too sure of all this stuff exactly, I believe that when each of us accesses the database that stores all the information in it, the database is locked while it is updated, which certainly means that traffic will slow down the speed at which requests can be processed. This difference is at least an order of magnitude larger than any difference in the time it takes the information to travel (assuming that your ISP is decent).
Incidentally, most of the hops are through routers, not hubs (which actually wouldn’t show up with a traceroute, IIRC) or computers.
Here is a traceroute and ping that I did just a little while ago. I’ll try to post back with results from some bizarre hour of the night to illustrate the difference in time and its insignificance.
(note cool way of getting time and date from my school’s server)
Anyways, that means that the packet spent 83.772 milliseconds in travel time, over all 19 hops. That is pretty insignificant when compared to how long it takes the server to actually process accessing the database.
Well, the board is now substantially slower than the last time I posted here. I mean, really friggin’ slow. I’d estimate that opening a thread takes twice as long as it did before. So let’s see what sort of traceroute and ping times we get:
Basically, now I’m going through 17 hops rather than 19, and it’s taking rather less time, only 38.995 milliseconds now. Yet downloading a page now is like waiting for the rain to erode Mt. Rushmore into four faceless blobs.
Oh, yeah, I should also mention that, as you can see from the difference in the two traceroutes, data can take different paths. If one path is broken, or a router is down, the data will be directed around it, at least for things in the middle. That’s the whole idea behind the internet.
Sorry for wasting everyone’s time with all this, but I get slightly annoyed when people post innaccurate descriptions of how the internet and this site work. That, and it’s a snow day today, due to the 16 or so inches of snow outside, so I’ve got some spare time.
waterj2-
Would you remind me how to execute a traceroute again?
Also, is it possible to control the path your connection makes to it’s destination? Say I want to send it through Florida as opposed to letting it do it itself. Is this possible?
Lots of snow in the Northeast. Kids are home because schools are closed. Parents of kids are home as well. Lots of people working from home as well. Less work being done. More straight dope being loaded.
Well, it’s now close to 3:00 AM here, and the board is nice and fast. Running a traceoute is as easy as typing “tracert boards.straightdope.com” into a DOS window. In my examples, I’m doing it from a Unix terminal, so the command is traceroute instead. I’ll look into the issue of sending data different ways and get back to you on that. Anyways, here we go again:
The average ping time has now gone all the way down to 36.720 milliseconds from a previous value of 38.995 milliseconds.
Or to analyze the whole traceroute, which fortunately follows the same path as before, with this afternoon on top and the middle of the night on the bottom:
1 128.213.7.254 (128.213.7.254) 0.582 ms 0.697 ms 0.531 ms
1 128.213.7.254 (128.213.7.254) 0.584 ms 0.513 ms 0.510 ms
2 spork (128.213.8.14) 1.013 ms 1.012 ms 1.012 ms
2 spork (128.213.8.14) 1.011 ms 1.038 ms 1.014 ms
3 vccfr4-113.its.rpi.edu (128.113.113.254) 1.649 ms 1.579 ms 1.721 ms
3 vccfr4-113.its.rpi.edu (128.113.113.254) 17.434 ms 1.660 ms 31.323 ms
4 vccfr6.nss.rpi.edu (128.113.39.251) 2.342 ms 2.265 ms 2.079 ms
4 vccfr6.nss.rpi.edu (128.113.39.251) 1.859 ms 1.992 ms 1.913 ms
5 at-gsr1-alb-0-3-renssel-1.appliedtheory.net (169.130.253.65) 4.930 ms 3.952 ms 3.147 ms
5 at-gsr1-alb-0-3-renssel-1.appliedtheory.net (169.130.253.65) 3.019 ms 3.308 ms 2.834 ms
6 at-gsr1-nyc-3-0-OC12.appliedtheory.net (169.130.3.29) 6.997 ms 6.058 ms 6.162 ms
6 at-gsr1-nyc-3-0-OC12.appliedtheory.net (169.130.3.29) 5.645 ms 6.525 ms 5.792 ms
7 sl-gw18-nyc-7-0.sprintlink.net (144.232.235.153) 12.793 ms 13.061 ms 13.630 ms
7 sl-gw18-nyc-7-0.sprintlink.net (144.232.235.153) 20.450 ms 21.264 ms 20.851 ms
8 sl-bb22-nyc-3-0.sprintlink.net (144.232.13.165) 21.490 ms 21.065 ms 21.365 ms
8 sl-bb22-nyc-3-0.sprintlink.net (144.232.13.165) 13.220 ms 12.356 ms 12.853 ms
9 ATM3-0.BR4.NYC9.ALTER.NET (137.39.52.21) 13.880 ms 13.757 ms 13.630 ms
9 ATM3-0.BR4.NYC9.ALTER.NET (137.39.52.21) 21.688 ms 21.265 ms 23.576 ms
10 0.so-6-3-0.XL1.NYC9.ALTER.NET (152.63.24.118) 13.291 ms 13.449 ms 13.319 ms
10 0.so-6-3-0.XL1.NYC9.ALTER.NET (152.63.24.118) 21.747 ms 21.088 ms 21.416 ms
11 0.so-7-0-0.XR1.NYC9.ALTER.NET (152.63.23.138) 13.145 ms 13.741 ms 13.988 ms
11 0.so-7-0-0.XR1.NYC9.ALTER.NET (152.63.23.138) 21.206 ms 21.452 ms 21.026 ms
12 0.so-4-0-0.TR1.NYC9.ALTER.NET (152.63.15.110) 21.210 ms 26.103 ms 21.458 ms
12 0.so-4-0-0.TR1.NYC9.ALTER.NET (152.63.15.110) 12.876 ms 13.674 ms 13.082 ms
13 125.at-5-0-0.TR1.CHI2.ALTER.NET (152.63.1.45) 30.075 ms 29.604 ms 30.027 ms
13 125.at-5-0-0.TR1.CHI2.ALTER.NET (152.63.1.45) 29.440 ms 29.973 ms 29.597 ms
14 197.ATM6-0.XR1.CHI6.ALTER.NET (152.63.65.81) 31.390 ms 32.239 ms 31.354 ms
14 197.ATM6-0.XR1.CHI6.ALTER.NET (152.63.65.81) 31.084 ms 31.257 ms 31.213 ms
15 191.ATM9-0-0.GW1.CHI6.ALTER.NET (146.188.208.73) 33.176 ms 33.051 ms 32.497 ms
15 191.ATM9-0-0.GW1.CHI6.ALTER.NET (146.188.208.73) 31.542 ms 32.206 ms 31.860 ms
16 node1.chicagoreader.com (63.97.40.1) 38.403 ms * 46.146 ms
16 node1.chicagoreader.com (63.97.40.1) 35.685 ms 36.072 ms 37.571 ms
17 node4.chicagoreader.com (63.97.40.4) 39.136 ms 45.098 ms 36.864 ms
17 node4.chicagoreader.com (63.97.40.4) 36.028 ms 36.095 ms 36.338 ms
Each hop along the way shows three times after it. If you compare each time from the afternoon to its mate from the middle of the night, I think we can safely eliminate trouble along the route as causing the problem. No step (except for some afternoon trouble at the Reader’s gateway) is significantly slower in the afternoon than now.
OK, I’ll shut up now. I think I’ve made my point. And, uh, sorry to those viewing this at less than 1024x768
I consistently get no more than 15 hops to the Home Page but as one of those is going to be about 3,500 miles perhaps it makes sense it’s less than some of you guys.
CnoteChris – This is quite a good product (lots and lots of info) if you want to do it simply. Have to pay for it after the trial period but I like it : http://www.neotrace.com/
Actually, the latest NeoTrace beta is free. I like the older 1.12 Shareware version best - its trial period doesn’t expire, and it has almost all functionality. More than the free beta, anyway.
Us mods are quite keen on NeoTrace. Can you guess why? 
Coldie, how are you finding the beta – any irritating crinkles to be ironed out or is it ready to go ?
The beta works fine, and looks supercool. Plus, it adds a menu bar button and right-click-menu option to IE5+ for quick scans. But the output is limited to a non-interactive trace list and a map, whereas the old 1.12 displayed much more, and allows you to hit nodes in the trace list for extra info.
waterj2, there are several reasons why the board can be slow; but is it necessary for me to go into detail for everyone of those possibilities for you? Here are some ideas:
There are times that their server is slow or overloaded.
If it is your first visit to the site or the required items are not in your browser’s cache (from
a previous visit) then they can take a little time to load.
If you are not using a modern modem then the loading of these items is may not be persistent and visible and may take longer.
Also some things that can effect the rate:
1.Your ISP’s (Internet Servive Provider) connection to the Internet
2.The segments of the Internet that you travel to reach your destination
3.The DNS (Domain Name Service) servers that you use
4.The Web Server that hosts the web site you are visiting.
If the board tests show that their slow connections and slow loading do not correspond to peak traffic times on the server and they have no reason to believe that the server has random ‘go slow’ periods then it could be things other than the server itself.
But I like the idea that all the computers & hubs have to be working right because I was addressing the Internet as a whole, (which is 75% of the reasons given above (1 thru 3)), a big bunch of computer servers & hubs, rather than just one tiny specific computer server, the board server, which is what you appear to be talking about & far be it from us to blame the Reader for the slowness.
handy, there are, of course, countless contributing factors to the board’s slowness. However, all other delays combined still pale in comparison to the delays on the boards.straightdope.com server itself, as has been demonstrated numerous times by the physical evidence, such as waterj2’s traceroutes. The server just plain has a hard time dealing with this much traffic and this many database accesses, and all the routers and hubs in the world can’t change that. To suggest otherwise, in the face of evidence, does nothing to further the cause of eradicating ignorance.
Thanks, Chronos , for telling it like it is.
I just spent from 10:00-11:10 PM Eastern Standard Time trying to accesss the message board here. Connections took many minutes for each step. I checked other message boards on my computer and they loaded in seconds.
All of a sudden, at 11:15 PM, or so, EST, the connections speeded up and loading was back to seconds.
Any person who has been reading these boards for even a few months knows that the poor overloaded server here is to blame.
My next question is this: what is to be done?
I am not very computer literate, but, is there one giant server or, rather, a group of servers which run this board? I know that a service such as AOL has millions of clients, so they have a huge server capacity. The SDMB has something over 10,000 potential clients, so, how big is/are our server(s)? We obviously don’t increase our server capacity fast enough to keep up with the increasing demand.
Is it just a matter of spending a few more bucks(how much is this?), or some other technical thingy that I might not understand? Is this on the order or needing to add capacity which costs a few hundred, a few thousand, or tens-of-thousands of dollars?
If it is question of large sums of money, why won’t the reader allow a token fee/donation/subsidy to be made by participants? Is it a legal problem? Ego problem? At some point, something has to be done. Can money cure the problem?
What is the relation between number of SDMB users and board speed? It is certainly not linear. My intuition tells me it’s like the “birthday paradox”, i.e if you put 23 people in a room, there’s a fifty-fifty chance that at least two of them share a birthday. Put 50 people in the room and it’s a virtual certainty.
Thanks!
Samclem:
In this thread (as well as related threads linked to in that one), the administration has declined to accept any kind of donations from its members.
Yeah, that’s kind of the idea. As the number of people increases, the number of things the server has to do simultaneously increases. I don’t know how the whole thing is set up, but there are a lot of things that have to happen in there. Given enough people, there’s inevitably going to be some retrieving things from the database from far flung spots on the drive, rather than just whatever happens to be stored in the cache. And it also has to write to the database, and actually serve the stuff out to everyone. All in all, a lot of places for stuff to get tied up. And of course the random ping comng in from some schmuck with nothing better to do.
And as far as I can tell, there is only one computer that holds the whole board. One, beautiful glorious computer with a worth far more than its weight in gold. Hopefully, there’s a backup somewhere.
Oh, and Chronos, actually, I think if the board’s traffic was routed through all the routers in the world, it probably would make a difference. That would be fun to do a traceroute on.
There used to be a link where you could see how many people were on the board when you came to it, but I can’t find it. If you can find it, water, you could compare speed with how many users are online.
Chronos, I think it would be rude of me to ask people about their server setup, I think of it as kinda private information, protected to ward off hackers, so I don’t ask about specifics…