UCSB Secure
is the name of the wireless network that members of the UC Santa Barbara community are encouraged to use. My impression of this network since it was rolled out several years ago has generally been that it's terrible, but that's a subjective impression, of course. This page aims to get some hard data about whether it is in fact terrible and, if so, just how terrible it is.
What's listed below are details (including transcripts) of comparatively long runs of the ping
command while I'm connected to the UCSB Secure network. ping
is not a perfect indicator of network performance and does not capture all of the subjective nuances of what's going on, but does give an index into network behavior (you can read more about this command on Wikipedia). I record the date, time, location on campus, length of the run for each run; provide a transcript of the run (I use the Linux script
command in a separate terminal emulator window to create these transcripts; the documentation for the script
command is here, if you want to read it), and provide some basic statistics for each transcript that I create. I try to remember to do this as often as possible, even before I start actively and directly using the network, and not just to document when there are problems (which is, after all, often enough. Still). I compress these files using bzip2 because I'm broke and paying out of pocket for my own web hosting (the University's technocrats, in their infinite wisdom, having decided several years ago to retire free web hosting for students).
My interpretation of the early and fragmentary data that I've gathered so far is that network performance on this network usually is, in fact, terrible, but this is of course a provisional conclusion that may change based on other data.
Date/ |
Run Length (msec); Link to Transcript | Packets Sent / Received (Packets Dropped: # / %) | rtt min/ping ) |
Notes |
---|---|---|---|---|
2015-10-28 |
2108498 ms; here | 2014 / 1903 (111 / 5.51%) | 4.122/ |
— |
2015-11-0211:31:55-08:00, CCS Trailer | 5034245 ms; here | 5016 / 3868 (1148 / 22.89%) | 4.298/ |
Network is essentially unusable, even for simple tasks. |
2015-11-04 |
7697197 ms; here | 6977 / 6918 (59 / 0.85%) | 3.736/ |
Alas, 51 of the 59 dropped packets are because I forgot to end the transcript before putting my computer to sleep and moving to a new location. |
2015-11-04 |
7697197 ms; here | 4485 / 4446 (39 / 0.87%) | 3.804/ |
Internet is sort of usable today. |
2015-11-09 |
4491228 ms; here | 7839 / 7599 (240 / 3.06%) | 3.838/ |
Problems are intermittent, but recur, and are severe when they're occurring. |
2015-11-16 |
7437253 ms; here | 7421 / 6923 (498 / 6.71%) | 4.081/ |
There's a genuine correlation observable today: When I'm not doing anything except pinging the network, the ping times are low, often in the range of a one-digit number of milliseconds. When I try to click on a link in a web page or have Thunderbird check my email, RTT time immediately increases by several hundred or several thousand times. That is, the network performs great, unless you're actually trying to use it, in which case it's terrible. |
2015-11-23 |
8346470 ms; here | 8834 / 8250 (584 / 6.61%) | 3.870/ |
Mostly usable most of the time, but measured jitter of over half a second points to widely variant experiences over that 2+ hours. Once again, as the day goes on, it gets worse. Once again, there were a lot of it works great when no one's using itmoments. |
2015-11-30 |
14783247 ms; here | 14762 / 14754 (8 / 0.05%) | 3.668/ |
Mostly usable most of the time. |
2015-12-02 |
20937840 ms; here | 20908 / 20759 (149 / 0.71%) | 3.792/ |
Another works fine, as long as you're not actually trying to use itkind of day. |
2015-12-10 |
14140750 ms; here | 14118 / 13934 (184 / 1.30%) | 3.727/ |
Mostly functional. |
2016-01-04 |
27307762 ms; here | 27263 / 27232 (31 / 0.11%) | 3.163/ |
Quite functional. |
2016-01-21 |
5505440 ms; here | 5499 / 5492 (7 / 0.13%) | 3.645/ |
Basically usable. |
2016-01-25 |
4048075 ms; here | 4042 / 3993 (49 / 1.21%) | 4.245/ |
Another if you're trying to route more than 64 bytes a second, good fucking luckday. |
2016-01-26 |
4978149 ms; here | 4970 / 4959 (11 / 0.22%) | 3.839/ |
— |
2016-01-28 |
8959739 ms; here | 8935 / 8558 (377 / 4.22%) | 3.895/ |
I hope there's a special place in hell for the people who get paid to administer this network. |
2016-01-28 |
5342307 ms; here | 5051 / 5040 (11 / 0.22%) | 3.772/ |
— |
2016-02-02 |
7559906 ms; here | 7547 / 7414 (133 / 1.76%) | 3.975/ |
A standard deviation of over two seconds is huge |
2016-02-02 |
5871460 ms; here | 5845 / 5842 (3 / 0.05%) | 3.698/ |
— |
2016-02-02 |
5871460 ms; here | 4093 / 3468 (625 / 15.27%) | 4.062/ |
And you thought RFC 1149 was an April Fools' Day joke. Guess again. |
2016-02-06 |
15909990 ms; here | 15885 / 15353 (532 / 3.35%) | 3.815/ |
Over 3% of packets being dropped and an average RTT of over 400ms is pretty bad. |
2016-02-11 |
12072979 ms; here | 12051 / 11322 (729 / 6.05%) | 3.949/ |
Not clear on what the difference is, from the user's perspective, between using a network administered by these people and using a network administered by people who are actually incompetent. |
2016-02-11 |
12072979 ms; here | 5376 / 5239 (137 / 2.55%) | 3.886/ |
Better than earlier today, but still fucking terrible. |
2016-02-16 |
5065899 ms; here | 5059 / 5050 (9 / 0.18%) | 4.600/ |
Why, it's almost usable. |
2016-02-16 |
6442107 ms; here | 6429 / 5805 (624 / 9.71%) | 4.153/ |
I hope there's a special place in hell for the people who administer this network. |
2016-02-18 |
12297818 ms; here | 12276 / 12040 (236 / 1.92%) | 3.930/ |
— |
2016-02-23 |
5585379 ms; here | 5575 / 5414 (161 / 2.89%) | 4.538/ |
— |
2016-02-25 |
8643280 ms; here | 8612 / 6736 (1876 / 21.78%) | 4.355/ |
You know how you can tell that the network is administered by people who, if they aren't completely fucking incompetent, might as well be? Because nearly two of every nine packets are dropped. |
2016-02-25 |
5205615 ms; here | 5198 / 5096 (102 / 1.96%) | 30.358/ |
— |
2016-03-01 |
6341892 ms; here | 6332 / 6328 (4 / 0.06%) | 12.962/ |
Slower than average, but more usable. |
2016-03-03 |
11395844 ms; here | 11373 / 10817 (556 / 4.89%) | 4.121/ |
Holy Christ. |
2016-03-03 |
4644310 ms; here | 4637 / 4614 (23 / 0.50%) | 4.843/ |
Better later in the afternoon and 50 feet away, I guess. |
2016-03-07 |
3912754 ms; here | 3908 / 3880 (28 / 0.72%) | 4.843/ |
Basically usable most of the time. |
2016-03-10 |
11824576 ms; here | 10949 / 10661 (288 / 2.63%) | 29.799/ |
0.5 ± 2.5 seconds for a packet trip, with 2.6% packet loss, is pretty horrible. |
2016-03-31 |
7203190 ms; here | 7192 / 6924 (268 / 3.73%) | 4.041/ |
Basically unusable with a nearly half-second average RTT and about 1 in 30 packets being dropped. |
2016-04-05 |
5056161 ms; here | 3918 / 3801 (117 / 2.99%) | 4.097/ |
Another half-second ± 2.5 seconds with about 3% dropped packetskind of day, just like normal. |
2016-04-07 |
6893736 ms; here | 6884 / 6871 (13 / 0.19%) | 4.156/ |
Basically usable today. |
2016-04-12 |
3161568 ms; here | 3157 / 3118 (39 / 1.24%) | 4.056/ |
Comparatively low levels of packet dropping don't make the connection actually usable. |
2016-04-12 |
10615125 ms; here | 10587 / 9184 (1403 / 13.25%) | 4.075/ |
Average ping time > 2.6 sec. ... for those not-quite-7-in-8 packets that actually manage to get back. |
2016-04-19 |
2592024 ms; here | 2588 / 2479 (109 / 4.21%) | 4.029/ |
Another day where trying to actually do anything on the Internet increases RTT by a thousandfold. |
2016-04-21 |
2659394 ms; here | 2654 / 2479 (175 / 6.59%) | 4.011/ |
Basically unusable most of the time. |
2016-04-28 |
14685390 ms; here | 14660 / 14408 (252 / 1.72%) | 4.093/ |
More usable than some times in the past; still, two-fifths of a second standard deviation in RTT points to a lot of jitter. |
2016-04-28 |
3924997 ms; here | 3917 / 3791 (126 / 3.22%) | 4.071/ |
Less usable than earlier this afternoon, presumably because I was actually trying to use it more of the time. |
2016-05-03 |
17094015 ms; here | 17064 / 16891 (173 / 1.01%) | 3.785/ |
— |
2016-05-05 |
7398320 ms; here | 7387 / 7336 (51 / 0.69%) | 4.292/ |
— |
2016-05-05 |
6857303 ms; here | 6845 / 6430 (415 / 6.06%) | 4.126/ |
Good thing I was just letting ping run while I was in class to measure this, as opposed to, say, actually using the Internet, which is clearly not the point of providing an ostensible Internet connection. |
2016-05-05 |
5293712 ms; here | 5288 / 5280 (8 / 0.15%) | 3.549/ |
Surprisingly usable. |
2016-05-08 |
37727232 ms; here | 37652 / 34171 (3481 / 9.25%) | 3.776/ |
Another works great, provided you’re not trying to route more than 64 bytes/seckind of day. |
2016-05-10 |
4591456 ms; here | 4576 / 2469 (2107 / 46.04%) | 4.318/ |
Yeah, totally useless. 3+ second average trip length with 6+ second standard deviation in trip length, and 46% of packets are being dropped. |
2016-05-10 |
8078011 ms; here | 8054 / 4910 (3144 / 39.04%) | 3.906/ |
Yeah, totally useless. |
2016-05-12 |
9423808 ms; here | 9400 / 8978 (422 / 4.49%) | 3.924/ |
Useless again. |
2016-05-12 |
5817624 ms; here | 5808 / 5805 (3 / 0.05%) | 4.584/ |
Useable, because I just left it running while class was going on, and wasn’t actually trying to use the Internet. The Internet is always usable when you’re not trying to use it. |
2016-05-12 |
3888730 ms; here | 3861 / 3481 (380 / 9.84%) | 4.166/ |
Useless, even though I’m hardly trying to use it this time. |
2016-05-17 |
3571854 ms; here | 3555 / 3064 (491 / 13.81%) | 4.424/ |
Totally unusable. |
2016-05-17 |
5701289 ms; here | 5692 / 5689 (3 / 0.05%) | 4.665/ |
Just left it running while sitting next to me in class. Demonstrates once again that the network is totally usable until you want to use it. |
2016-05-17 |
5334523 ms; here | 5224 / 4512 (712 / 13.63%) | 3.924/ |
Once again, totally unusable. |
2016-05-19 |
9016697 ms; here | 9002 / 8908 (94 / 1.04%) | 4.203/ |
Sort of usable some of the time. |
2016-05-24 |
4681324 ms; here | 3858 / 2462 (1396 / 36.18%) | 4.072/ |
Totally unusable. |
2016-05-24 |
5263517 ms; here | 5247 / 2761 (2486 / 47.38%) | 4.123/ |
Totally unusable. |
2016-05-25 |
6284931 ms; here | 6276 / 5720 (556 / 8.86%) | 3.715/ |
Totally unusable. |
2016-05-25 |
11061916 ms; here | 8856 / 8749 (107 / 1.21%) | 3.692/ |
Didn’t actually try to use it—I just left it running while engaged in something else—so of course it worked fine. |
2016-05-25 |
6984252 ms; here | 6932 / 6659 (273 / 3.94%) | 4.023/ |
Almost usable. |
2016-05-26 |
6709027 ms; here | 6697 / 6370 (327 / 4.88%) | 3.955/ |
— |
2016-05-26 |
5124459 ms; here | 5117 / 5116 (1 / 0.02%) | 4.743/ |
Once again, the network performs just great, provided that you’re not asking it to transmit more than 64 bytes per second. |
2016-05-27 |
3725907 ms; here | 3720 / 3716 (4 / 0.11%) | 4.585/ |
It works great on the Friday of a three-day holiday weekend, provided you’re not asking it to transmit more than 64 bytes per second because you just left it running while you were out of your office for an hour. |
2016-05-28 |
24229197 ms; here | 24164 / 17307 (6857 / 28.38%) | 4.107/ |
Yeah, totally useless. |
2016-05-31 |
3520265 ms; here | 3516 / 3489 (27 / 0.77%) | 4.038/ |
Basically usable. |
2016-06-02 |
9809828 ms; here | 9787 / 9198 (589 / 6.02%) | 3.767/ |
Incredibly slow, when it works. |