aboutsummaryrefslogtreecommitdiff
path: root/spec/bandwidth-file-spec/scaling-bandwidths.md
diff options
context:
space:
mode:
authorNick Mathewson <nickm@torproject.org>2023-10-13 18:00:42 -0400
committerNick Mathewson <nickm@torproject.org>2023-10-13 18:00:42 -0400
commitf79272ef1f774b3788b74a3fe4fef75095dfae06 (patch)
tree8f47bebaa06c444f632bf8c4afbd793c4972a27d /spec/bandwidth-file-spec/scaling-bandwidths.md
parentfa014ec90411fd754dd257d04afa1a953e15bf31 (diff)
downloadtorspec-f79272ef1f774b3788b74a3fe4fef75095dfae06.tar.gz
torspec-f79272ef1f774b3788b74a3fe4fef75095dfae06.zip
Run markdownlint --fix on spec.
Diffstat (limited to 'spec/bandwidth-file-spec/scaling-bandwidths.md')
-rw-r--r--spec/bandwidth-file-spec/scaling-bandwidths.md6
1 files changed, 5 insertions, 1 deletions
diff --git a/spec/bandwidth-file-spec/scaling-bandwidths.md b/spec/bandwidth-file-spec/scaling-bandwidths.md
index 8ae80c9..32fcb83 100644
--- a/spec/bandwidth-file-spec/scaling-bandwidths.md
+++ b/spec/bandwidth-file-spec/scaling-bandwidths.md
@@ -1,7 +1,9 @@
<a id="bandwidth-file-spec.txt-B"></a>
+
# Scaling bandwidths
<a id="bandwidth-file-spec.txt-B.1"></a>
+
## Scaling requirements
```text
@@ -18,6 +20,7 @@ torflow and sbws, because their measured bandwidths are similar
enough already.
<a id="bandwidth-file-spec.txt-B.2"></a>
+
## A linear scaling method
If scaling is required, here is a simple linear bandwidth scaling
@@ -47,6 +50,7 @@ Now, the total scaled bandwidth in the upcoming vote is
approximately equal to the quota.
<a id="bandwidth-file-spec.txt-B.3"></a>
+
## Quota changes
If all generators are using scaling, the quota can be gradually
@@ -56,6 +60,7 @@ consensus diffs and compressed consensuses. But if the relay
quota is too small, some relays may be over- or under-weighted.
<a id="bandwidth-file-spec.txt-B.4"></a>
+
## Torflow aggregation
Torflow implements two methods to compute the bandwidth values from the
@@ -125,4 +130,3 @@ r_strm_i = bw_i / bw_avg_strm
<<In this way, the resulting network status consensus bandwidth
values are effectively re-weighted proportional to how much faster
the node was as compared to the rest of the network.>>
-