Internet Speed Tests

Ya you can use the included developer tools in most browsers for many statistics, including DNS lookup and download times, e.g.

On a sunny day (Sat, 29 Jan 2022 18:46:55 -0800 (PST)) it happened Rick C snipped-for-privacy@gmail.com wrote in snipped-for-privacy@googlegroups.com:

If I want to know if my internet connection is normal speed I simply type: ping 8.8.8.8 in a Linux terminal. That is the google nameserver ~# ping 8.8.8.8 PING 8.8.8.8 (8.8.8.8) 56(84) bytes of data.

64 bytes from 8.8.8.8: icmp_req=1 ttl=57 time=34.9 ms

Any slower and my 4G is not working,,,

You can also download a file with "wget" and it will show the speed for that site:

~# wget

07:26:51-- panteltje.com (panteltje.com)... 92.205.4.14 Connecting to panteltje.com (panteltje.com)|92.205.4.14|:80... connected. HTTP request sent, awaiting response... 200 OK Length: 2855 (2.8K) [text/html] Saving to: 'index.html'

100%[======================================================================================================================================================================================>] 2,855 --.-K/s in 0s

2022-01-30 07:26:52 (71.6 MB/s) - 'index.html' saved [2855/2855]

-------------------^^^^^^^^^^^^

Better use a longer file else the MB/s may not be real (cached somewhere perhaps).

There are websites that will show you your own IP address, google for it.

If all else fails maybe buy a SpaceX satellite terminal?

It is impractical to make such a reliable test. Much of its results can depend on the user testing it, like which dns server they use; a typical website accesses a lot of various IP addresses. Most likely what is annoying you are the ad related accesses, if some google or whatever central thing is slow you are made to wait for it - sometimes for a very long time. Then some addresses (IP addresses) are dynamically moved from one location to another, it is quite a mess really. The tests available you are aware of do what is practical to do; you can see how fast down and upload work, how they reach full speed (for tcp it is mandatory to have a slow start transmitting, not necessarily slow enough to be easily noticeable though) and that is it. You can change test server location, various destinations are routed differently and some are slower than others.

Most browser-based speed tests report ping or latency time. A specific web site could be anything over that.

The M-Lab test (the google default) shows me 135/40/7ms, Comcast cable.

I guess my point is timing the download of a large file is not a good measure of much of anything. Web sites consist of many files, often none of them large. The time it takes for a web page to be viewable depends on the time it takes for the entire protocol of loading the HTML, requesting the various files specified there, then loading the various files specified there, and so on. The observed delays in the initial display in the browser of web pages is often much, much longer than any transmission time of the files in a page. I see large variations between different Internet providers, so it's not my PC or browser.

I did fix an issue I periodically have in Virginia. Seems my router has developed some problems. I replaced it and am getting much better results now. Ping times in the teens.

Each of those those various files will also have an associated domain name lookup, so changing to a domain name server that responds more quickly can make a big difference.

Another factor which can sometimes make websites seem very slow is that occasionally IPv6 connectivity is broken. Most browsers nowadays will use the IPv6 address first and if it doesn't respond within a certain time will then try IPv4. If the default gateway is announcing an IPv6 route which doesn't work then everything gets very slow.

John

It is still not a bad proxy for internet speed.

The other one is ping time to the server which gives you a good idea of the round trip time for the minimal short message.

Often for web pages your delays are due to a dodgy slow DNS lookup or some turgid lethargic script on an overloaded web server.

Pick a reasonably static website and check the time to render it. Or image it to your hard disk. Various spiderlike software exists to do this - the owner of the website may take exception if you do it too often or for large chunks of their content.

Most websites these days come up so quickly on a fast line that you may need to time it in software rather than manually. The only ones that don't are corporates with 100MB video files on the landing page :(

Many sites don't like it if your IP address changes mid session.

That could very well be due to the "slow start" architecture in TCP/IP. TCP connections deliberately start out with a relatively modest "receive window" size, and then ramp up the window size once the sending and receiving systems have exchanged enough data for a meaningful evaluation of the actual performance.

The intent here (as I recall it) is to make sure that the sending system (server) doesn't shove out a huge glob of data faster than the receiving system can actually receive and process it. To do so, would either fill up buffers in the intervening network switches/routers, or cause packets to be dropped when the buffers overflow... and these things have a bad effect on network performance and reliability.

This means that the initial connection for a web page can be a bit slow getting started. Web browsers (and web servers) these days try to overcome this by a couple of tricks:

- Most servers can handle multiple requests, one after another, on a single TCP connection. Web browsers take advantage of this by "pipelining" several such requests... once they receive the main page from the server and start to parse it, they fire off additional requests for the other resources mentioned in that page which are on the same server. This allows the incoming server-to-client connection to get past the slow-start stage and deliver content at full speed.

- Web browsers will usually make connection to different servers in parallel.

There's so much variation in how web pages are organized (number of resources they fetch, sizes of those resources, where they are fetched from) that it's awfully difficult to develop an honest apples-and-apples comparison.

This problem is made even worse by the fact that many web servers are "virtual" - that is, there are numerous "clones" of a given content distribution server scattered around the net, and DNS or other tricks are used to route your request to the "best" (closest, fastest) server at any given instant in time.

As a result, if you run the same test twice a few minutes apart, or switch provider networks to do a comparison, you may end up fetching web content from a completely different set of servers.

If slow web surfing were the result of inherent mechanisms in the TCP/IP protocol, this would be observed on every computer, on every network through every ISP to every web site. I don't see this. I see wide variations in both raw speed and the time to get a transfer started. When there is a bottle neck in the system the speed tests often start at some very low rate and slowly ramp up to a more reasonable speed. On faster networks the ramp up is only a small amount with the initial transfer speed rather fast.

Ad servers are slow partly because they take time to /auction/ your eyeballs to the highest bidder. Yes, they pass info to many of their customers, so their customers can decide how valuable (or not) you are to them.

NoScript and AdBlock are necessary when browsing the web. Carl Sagan foresaw the necessity for them in his novel Contact, albeit with TV advertising rather than the web.

Probably all of your suspicions of what they do are correct and likely we can't suspect enough :-). Things become messier by the day, lately the ISP-s (or some entity behind them) join in; for example, since may be a month if I start facebook-browsing (I am not a very active facebooker but some days I spend 5-10 minutes, mainly looking at posts to local village groups and the poster's profiles) after only 2-3 minutes of active browsing certain parts become non-responsive (not the main page, just what it references, photos, posts, menus etc.). And this is not facebook's fault, if I log in via TOR things work just fine, so it is either the local ISP or something between them and facebook, who knows. My guess is the local ISP get too much facebook traffic (all those kids with their phones) and limit it but it is as good as anybody's guess.

Anyway, I don't even try to guess who is doing what on the web. I don't switch scripts off or use adblockers, I guess I don't waste that much time browsing. Mostly the BBC website, football scores etc., I don't know how much of these will work if I block the ads, so far it is tolerable for me.

browsing facebook is a drop in the ocean compared to watching a movie

everything still works, just a million times better

install something like adblock+ and you'll never go back

Try adblock; you can always uninstall/disable it.

NoScript is more for the tin-foil hat brigade, and requires nursing.

Some ISPs (IIRC Comcast? Verizon?) use deep packet inspection /and modification/ to modify web pages and insert /their/ chosen adverts.

Why do you think many pages have farcebook and twatter (and other) logos on them? Basically it is so those companies can tell which pages your browser has visited. That takes time to complete.

I wonder, do you think this could have anything to do with the fact that I am using dial up?

Except that it prevents you from accessing some sites. I suppose most of those are good riddance.

very few sites and most of the blocks quickly get circumvented by the adblockers

I don't visit very many sites and they repeatedly block my access. I use google news and it is common for the referred web sites to block ad blockers by putting up a overlay telling you to turn off your ad blocker. Some you can tell to simply go away. Others are persistent and you can't view the page. Google news has a feature to let you block seeing the headlines of a site, so the ones which are persistent get removed from my view. Fox News recently bit the dust this way and that's one I would like to read. They may be a bit extreme, but not always and it is good to hear from all perspectives. I think I blocked Reuters as well because of their ad blocker block.

I use several browsers and have uBlock on this one (Firefox). I seem to be using AdBlock on Chrome. I also use a Comodo variant of the Chrome browser with their own tool that is more about security, but also blocks ads I believe. Then I use an AVG browser which of course, blocks ads and has security. I think I have the most trouble with AdBlock on Chrome, but that's probably because I view news on Chrome where most of the ads show up.

Ouch, I must have missed this fact. I assumed you have some sort of broadband. At dial up speeds I imagine a lot can go wrong, sites are widely optimized/tested at much higher speeds than that. Browser can time out on certain pages, drop connections etc. I have seen that when my phone internet access gets down to 64 kbps because I have my limits exhausted, things barely work really.

Sorry, I should have used a smiley. ;)