https://github.com/nginx-modules/ngx_http_tls_dyn_size

Optimizing TLS over TCP to reduce latency for NGINX
https://github.com/nginx-modules/ngx_http_tls_dyn_size

cloudflare dynamic http2 https nginx optimization segment tcp tls

Last synced: 12 days ago
JSON representation

Optimizing TLS over TCP to reduce latency for NGINX

• Host: GitHub
• URL: https://github.com/nginx-modules/ngx_http_tls_dyn_size
• Owner: nginx-modules
• Created: 2016-12-01T17:14:58.000Z (almost 8 years ago)
• Default Branch: master
• Last Pushed: 2024-05-04T19:16:00.000Z (6 months ago)
• Last Synced: 2024-08-01T19:33:05.705Z (3 months ago)
• Topics: cloudflare, dynamic, http2, https, nginx, optimization, segment, tcp, tls
• Homepage: https://blog.cloudflare.com/optimizing-tls-over-tcp-to-reduce-latency/
• Size: 19.5 KB
• Stars: 30
• Watchers: 6
• Forks: 7
• Open Issues: 1
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# Optimizing TLS over TCP to reduce latency for NGINX

* [`nginx__dynamic_tls_records`](https://github.com/cloudflare/sslconfig/blob/3e45b99/patches/)

* [`Optimizing HTTP/2 prioritization with BBR and tcp_notsent_lowat`](https://blog.cloudflare.com/http-2-prioritization-with-nginx/)

### What we do now

We use a static record size of 4K.

This gives a good balance of latency and throughput.

#### Configuration

*Example*

```nginx

http {

  ssl_dyn_rec_enable on;

}

```

#### Optimize latency

By initialy sending small (1 TCP segment) sized records,

we are able to avoid HoL blocking of the first byte.

This means TTFB is sometime lower by a whole RTT.

#### Optimizing throughput

By sending increasingly larger records later in the connection,

when HoL is not a problem, we reduce the overhead of TLS record

(29 bytes per record with GCM/CHACHA-POLY).

#### Logic

Start each connection with small records

(1369 byte default, change with `ssl_dyn_rec_size_lo`).

After a given number of records (40, change with `ssl_dyn_rec_threshold`)

start sending larger records (4229, `ssl_dyn_rec_size_hi`).

Eventually after the same number of records,

start sending the largest records (`ssl_buffer_size`).

In case the connection idles for a given amount of time

(1s, `ssl_dyn_rec_timeout`), the process repeats itself

(i.e. begin sending small records again).

### Configuration directives

#### ssl_dyn_rec_enable

* **syntax**: `ssl_dyn_rec_enable bool`

* **default**: `off`

* **context**: `http`, `server`

#### ssl_dyn_rec_timeout

* **syntax**: `ssl_dyn_rec_timeout number`

* **default**: `1000`

* **context**: `http`, `server`

We want the initial records to fit into one TCP segment

so we don't get TCP HoL blocking due to TCP Slow Start.

A connection always starts with small records, but after

a given amount of records sent, we make the records larger

to reduce header overhead.

After a connection has idled for a given timeout, begin

the process from the start. The actual parameters are

configurable. If `ssl_dyn_rec_timeout` is `0`, we assume `ssl_dyn_rec` is `off`.

#### ssl_dyn_rec_size_lo

* **syntax**: `ssl_dyn_rec_size_lo number`

* **default**: `1369`

* **context**: `http`, `server`

Default sizes for the dynamic record sizes are defined to fit maximal

TLS + IPv6 overhead in a single TCP segment for lo and 3 segments for hi:

1369 = 1500 - 40 (IP) - 20 (TCP) - 10 (Time) - 61 (Max TLS overhead)

#### ssl_dyn_rec_size_hi

* **syntax**: `ssl_dyn_rec_size_hi number`

* **default**: `4229`

* **context**: `http`, `server`

4229 = (1500 - 40 - 20 - 10) * 3  - 61

#### ssl_dyn_rec_threshold

* **syntax**: `ssl_dyn_rec_threshold number`

* **default**: `40`

* **context**: `http`, `server`

### License

* [Cloudflare](https://github.com/cloudflare), [Vlad Krasnov](https://github.com/vkrasnov)