Skip to content

Commit 0c0880c

Browse files
tohojotohojo
committed
Initial import.
This is just importing everything from the bufferbloat wiki db dump. The files in content/wiki are generated by the convert.py script in export/ which also contains the source csv files. The layout/ is from blog.cerowrt.org but adjusted to work for the wiki/ section.
0 parents  commit 0c0880c

File tree

282 files changed

+54619
-0
lines changed

Some content is hidden

Large Commits have some content hidden by default. Use the searchbox below for content that may be hidden.

282 files changed

+54619
-0
lines changed

‎.gitignore

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1 @@
1+
public

‎config.yaml

Lines changed: 3 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,3 @@
1+
baseurl: "http://archive.bufferbloat.net/"
2+
languageCode: "en-us"
3+
title: "Bufferbloat.net archive"

‎content/wiki/172160012_allocation.md

Lines changed: 15 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,15 @@
1+
2+
---
3+
title: 172160012_allocation
4+
date: 2011-07-09T10:05:06
5+
lastmod: 2011-07-09T10:05:06
6+
---
7+
172.16.0.0/12 Allocation
8+
========================
9+
10+
We kind of hope to be fiddling with vpn technology at some point, to
11+
connect widely disconnected researchers together, much like how the
12+
early ARPA-net did the same for universities around the United States
13+
back in the early 1970s.
14+
15+
See <link>bloat:BANA</link> for more details.

‎content/wiki/338-11_tests.md

Lines changed: 81 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,81 @@
1+
2+
---
3+
title: 338-11_tests
4+
date: 2012-07-12T16:42:27
5+
lastmod: 2012-07-12T16:42:28
6+
---
7+
338-11 tests
8+
============
9+
10+
m@cruithne:~/3.3.8-11$ netperf -H 172.25.45.1
11+
MIGRATED TCP STREAM TEST from 0.0.0.0 (0.0.0.0) port 0 AF_INET to 172.25.45.1 () port 0 AF_INET
12+
Recv Send Send
13+
Socket Socket Message Elapsed
14+
Size Size Size Time Throughput
15+
bytes bytes bytes secs. 10^6bits/sec
16+
17+
87380 65536 65536 10.06 37.80
18+
m@cruithne:~/3.3.8-11$ netperf -l 60 -H 172.25.45.1 &
19+
[1] 19912
20+
m@cruithne:~/3.3.8-11$ MIGRATED TCP STREAM TEST from 0.0.0.0 (0.0.0.0) port 0 AF_INET to 172.25.45.1 () port 0 AF_INET
21+
netperf -l 60 -H 172.25.45.1 &
22+
[2] 19913
23+
m@cruithne:~/3.3.8-11$ MIGRATED TCP STREAM TEST from 0.0.0.0 (0.0.0.0) port 0 AF_INET to 172.25.45.1 () port 0 AF_INET
24+
netperf -l 60 -H 172.25.45.1 &
25+
[3] 19914
26+
m@cruithne:~/3.3.8-11$ MIGRATED TCP STREAM TEST from 0.0.0.0 (0.0.0.0) port 0 AF_INET to 172.25.45.1 () port 0 AF_INET
27+
netperf -l 60 -H 172.25.45.1 &
28+
[4] 19915
29+
m@cruithne:~/3.3.8-11$ MIGRATED TCP STREAM TEST from 0.0.0.0 (0.0.0.0) port 0 AF_INET to 172.25.45.1 () port 0 AF_INET
30+
Recv Send Send
31+
Socket Socket Message Elapsed
32+
Size Size Size Time Throughput
33+
bytes bytes bytes secs. 10^6bits/sec
34+
35+
87380 65536 65536 60.45 18.32
36+
Recv Send Send
37+
Socket Socket Message Elapsed
38+
Size Size Size Time Throughput
39+
bytes bytes bytes secs. 10^6bits/sec
40+
41+
87380 65536 65536 60.23 16.76
42+
Recv Send Send
43+
Socket Socket Message Elapsed
44+
Size Size Size Time Throughput
45+
bytes bytes bytes secs. 10^6bits/sec
46+
47+
87380 65536 65536 60.17 16.35
48+
Recv Send Send
49+
Socket Socket Message Elapsed
50+
Size Size Size Time Throughput
51+
bytes bytes bytes secs. 10^6bits/sec
52+
53+
87380 65536 65536 60.04 16.50
54+
55+
root@village:~# tc -s qdisc show dev sw10
56+
qdisc mq 1: root
57+
Sent 8347990 bytes 100563 pkt (dropped 121777, overlimits 0 requeues 11389)
58+
backlog 39006b 591p requeues 11389
59+
qdisc fq_codel 10: parent 1:1 limit 100p flows 1024 quantum 1514 target 5.0ms interval 100.0ms
60+
Sent 0 bytes 0 pkt (dropped 0, overlimits 0 requeues 0)
61+
backlog 0b 0p requeues 0
62+
maxpacket 256 drop_overlimit 0 new_flow_count 0 ecn_mark 0
63+
new_flows_len 0 old_flows_len 0
64+
qdisc fq_codel 20: parent 1:2 limit 1200p flows 1024 quantum 1514 target 13.0ms interval 100.0ms ecn
65+
Sent 864 bytes 8 pkt (dropped 0, overlimits 0 requeues 0)
66+
backlog 0b 0p requeues 0
67+
maxpacket 256 drop_overlimit 0 new_flow_count 0 ecn_mark 0
68+
new_flows_len 0 old_flows_len 0
69+
qdisc fq_codel 30: parent 1:3 limit 1200p flows 1024 quantum 1514 target 13.0ms interval 100.0ms ecn
70+
Sent 6963197 bytes 98910 pkt (dropped 121777, overlimits 0 requeues 11253)
71+
^^^^^ ^^^^^ WTF?
72+
backlog 39006b 591p requeues 11253
73+
maxpacket 1514 drop_overlimit 0 new_flow_count 952 ecn_mark 33
74+
new_flows_len 0 old_flows_len 6
75+
qdisc fq_codel 40: parent 1:4 limit 1000p flows 1024 quantum 1514 target 50.0ms interval 100.0ms
76+
Sent 1383929 bytes 1645 pkt (dropped 0, overlimits 0 requeues 136)
77+
backlog 0b 0p requeues 136
78+
maxpacket 1514 drop_overlimit 0 new_flow_count 184 ecn_mark 0
79+
new_flows_len 0 old_flows_len 0
80+
81+
doing TCP\_MAERTS instead...

‎content/wiki/ACK_prioritization.md

Lines changed: 42 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,42 @@
1+
2+
---
3+
title: ACK_prioritization
4+
date: 2011-06-20T05:43:13
5+
lastmod: 2011-06-20T05:43:32
6+
---
7+
ACK prioritization
8+
==================
9+
10+
In today's asymmetric internet environment up and download ratios can be
11+
as poor as 1 to 11. While this works well for situations where downloads
12+
predominate, it can cause terrible side effects on an otherwise healthy
13+
network when an upload saturates the connection, starving the downloads
14+
of needed ACK packets.
15+
16+
Ratios worse than 11 to 1 appear to touch closely to the theoretical
17+
limit of \~23 to 1 that TCP/IP invokes in it's control stream.\
18+
This is why, in part, worse ratios haven't appeared in vendor offerings,
19+
as downloads are effectively throttled by upload bandwidth.
20+
21+
Since ack packets are very small (less than 72 bytes in ipv4 and less
22+
than 140 bytes in IPv6, depending on encapsulation), shaping methods
23+
that depend more on packets than bytes tend to suffer. 23 (ipv4) ack
24+
packets can fit into the same amount of buffer space as a single upload
25+
packet.
26+
27+
It is helpful from a policing perspective to look more at bytes than
28+
packets, for uploads in an asymmetric network, to determine what packets
29+
to best 'shoot'. Even then, the side effects of shooting an upload
30+
packet instead of 23 ack packets tend to be more beneficial than often
31+
realized.
32+
33+
Lastly, no shaping system takes into account the up/download ratio in
34+
its decision making particularly well.
35+
36+
Several tries exist - notably, wondershaper, which pioneered the concept
37+
of doing ACK prioritization for interactive ssh traffic back in 2002.
38+
39+
- Note also that IPv6 changes the ack equation markedly - an
40+
encapsulated IPv6 6in4 ack packet can be as large as 140
41+
(reference needed)
42+

‎content/wiki/AHCP.md

Lines changed: 26 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,26 @@
1+
2+
---
3+
title: AHCP
4+
date: 2012-08-17T14:20:22
5+
lastmod: 2012-08-17T14:20:22
6+
---
7+
AHCP
8+
====
9+
10+
ahcp is the [ad-hoc configuration
11+
protocol](http://www.pps.univ-paris-diderot.fr/~jch/software/ahcp/)
12+
created by [Juliusz
13+
Chroboczek](http://www.pps.univ-paris-diderot.fr/~jch/) .
14+
15+
It uses link-local ipv6 multicast to distribute ipv4 and ipv6 addresses,
16+
ntp and name server information. It differs from dhcp and dhcpv6 in that
17+
it:
18+
19+
Uses one packet to distribute ipv4 and ipv6 information (dhcp and dhcpv6
20+
require separate and more packets each)\
21+
Distributes point to point addresses (/32 ipv4 and /128 ipv6)\
22+
Lacks the backward compatability cruft in dhcp and dhcpv6
23+
24+
For more information, including the current specification and rfc,
25+
please see the [ahcp web
26+
site](http://www.pps.univ-paris-diderot.fr/~jch/software/ahcp/) .

‎content/wiki/Advice_for_ISPS.md

Lines changed: 20 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,20 @@
1+
2+
---
3+
title: Advice_for_ISPS
4+
date: 2015-06-18T10:51:28
5+
lastmod: 2015-06-18T10:51:28
6+
---
7+
Advice for ISPs
8+
===============
9+
10+
DO:
11+
---
12+
13+
- Measure
14+
-
15+
16+
Don't
17+
-----
18+
19+
- overuse policing
20+
-

‎content/wiki/Analysis_of_existing_firewalls_and_shapers.md

Lines changed: 49 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,49 @@
1+
2+
---
3+
title: Analysis_of_existing_firewalls_and_shapers
4+
date: 2011-06-20T05:28:52
5+
lastmod: 2011-06-20T05:28:52
6+
---
7+
Analysis of existing firewall and shaper scripts
8+
------------------------------------------------
9+
10+
TBD: I have examples of many of these types of shapers that I plan to go
11+
into more detail with shortly.
12+
13+
### Openwrt firewall + QoS rules
14+
15+
Openwrt ties it's firewalling and QoS code closely together, using a
16+
combination of 'qos-scripts' and firewall rules in a somewhat easy to
17+
read format in /etc/config/qos and /etc/config/firewall, generating
18+
complex rules as a result. It also defaults to TCP Westwood+ which has
19+
interesting interactions with other TCP traffic when a proxy is used.
20+
21+
### Gargoyle
22+
23+
Gargoyle (A fork of openwrt)
24+
25+
### ufw
26+
27+
### Wondershaper
28+
29+
Wondershaper's big claim to fame was it's simplicity. It pioneered
30+
<link>ACK prioritization</link> for ssh traffic, and did its work in
31+
only 4 TC rules.
32+
33+
### Nanog
34+
35+
### Airmax
36+
37+
Used by ubiquity in their line of high performance wireless routers,
38+
this consists of a lot of very hard to parse tc rules that work magic
39+
for fairness across a wireless network
40+
41+
### Adsl-Shaper
42+
43+
### Shorewall
44+
45+
### Linux voip server example
46+
47+
### Linux servers
48+
49+
### Linux desktops

‎content/wiki/App-brainstorm.md

Lines changed: 66 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,66 @@
1+
2+
---
3+
title: App-brainstorm
4+
date: 2011-07-14T05:44:46
5+
lastmod: 2011-07-14T05:44:46
6+
---
7+
Application Brainstorm
8+
======================
9+
10+
Different types of users: home users and researchers.
11+
12+
Toolkit for building applications? Could have a router platform for
13+
allowing other users to develop without knowledge of or access to the
14+
low-level Apis, similar to how people develop applications for iPhone.
15+
16+
Seattle, Scriptroute, etc. Compare to the POX design.
17+
18+
First step could be to start with a limited problem domain and design
19+
the API around it.
20+
21+
Privacy of data collected. Guarantee privacy of collected information
22+
with information flow control.
23+
24+
Control: home automation. Does an openwrt router support zwave or any
25+
home automation type thing? Could couple with home automation? Use
26+
network router to track human activity and then couple with a control
27+
system.
28+
29+
Take Keith's findings about what network features are useful for
30+
determining human behavior and then use that as a basis for an API.
31+
32+
Two parts to the API: monitoring and control. What kind of questions to
33+
ask for each type?
34+
35+
Is it worth doing a survey to figure out the types of things people
36+
would like to know from their home network, and the types of things they
37+
would like to control. Conduct a broader survey?
38+
39+
Would certain types of queries impose storage requirements?
40+
41+
Version control for router configurations. Provide for the users in the
42+
cloud. Configuration management for home networks?
43+
44+
USB stick for extra storage. Caching for updates. Caching for improving
45+
performance of different applications.
46+
47+
Simultaneous high bandwidth downloads introducing interference? Time
48+
shifting big downloads?
49+
50+
What to cache? Advantages to caching on the gateway versus caching on
51+
the browser?
52+
53+
What problems will go away? Latency is pretty fundamental but doesn't
54+
apply to all applications. Throughout may be a longer-term issue.
55+
56+
Questions for netgear?
57+
58+
Diagnostic tool for the web interface to let users run an instant
59+
diagnostic
60+
61+
1\. End users:
62+
63+
2\. Researchers:\
64+
Building better tools\
65+
Creating interfaces\
66+
Using the platform as a tool to monitor human behavior

‎content/wiki/Architecture_Notes.md

Lines changed: 16 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,16 @@
1+
2+
---
3+
title: Architecture_Notes
4+
date: 2011-06-04T14:52:47
5+
lastmod: 2011-06-24T13:31:46
6+
---
7+
Architecture Notes
8+
==================
9+
10+
http://www.mips.com/media/files/MD00565-2B-MIPS32-QRC-01.01.pdf
11+
12+
<link>OCEAN CITY FAQ</link>
13+
14+
<link>scratch</link>
15+
16+
<link>OCEAN CITY INSTALLATION GUIDE</link>

‎content/wiki/Atom.md

Lines changed: 34 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,34 @@
1+
2+
---
3+
title: Atom
4+
date: 2011-04-07T18:17:01
5+
lastmod: 2011-04-07T18:17:01
6+
---
7+
Atom and BISMark
8+
================
9+
10+
You will need to download the latest [images are
11+
here](http://huchra.bufferbloat.net/atom)
12+
13+
And then install.
14+
15+
These are the wndr instructions, this needs to be rewritten for the
16+
atom:
17+
18+
[Installation
19+
instructions](http://wiki.openwrt.org/toh/netgear/wndr3700#oem.easy.installation)\
20+
Specific <link>Flashing instruction for Mac</link>.
21+
22+
If you need additional packages, you will need to remove /etc/opkg/\*\
23+
then edit the /etc/opkg.conf file to point to either
24+
http://huchra.bufferbloat.net/atom instead of what it points to
25+
currently, then do a
26+
27+
`opkg update
28+
opkg list
29+
opkg install your_package`
30+
31+
There will be a special <link>bismark feed</link> at some point
32+
33+
Please note that huchra will move to a saner name at some point in the
34+
near future, and we hope to make these defaults somehow.

0 commit comments

Comments
 (0)