Scaling

The Luna-Streams-Retranslator service scaling has some peculiarities.

The proper way to use multiple instances of Luna-Streams-Retranslator is to split requests to service by using different ranges of stream_id for each instance.

Here and below all examples will use configuration of nginx.

Warning

Load balancer listening address and port must match the ones from EXTERNAL_LUNA_STREAMS_HLS_RETRANSMISSION_ADDRESS setting.

Foreword

Each stream has its unique id. Stream id represents uuid4. It is randomly generated when stream is created. So it assumes that they are distributed evenly.

Examples of split uuid4 by ranges:

• two parts:

  • from 00000000-0000-4000-8000-000000000000 to 80000000-0000-4000-8000-000000000000

  • from 90000000-0000-4000-8000-000000000000 to ffffffff-ffff-4fff-bfff-ffffffffffff

• four parts:

  • from 00000000-0000-4000-8000-000000000000 to 30000000-0000-4000-8000-000000000000

  • from 40000000-0000-4000-8000-000000000000 to 80000000-0000-4000-8000-000000000000

  • from 90000000-0000-4000-8000-000000000000 to a0000000-0000-4000-8000-000000000000

  • from c0000000-0000-4000-8000-000000000000 to ffffffff-ffff-4fff-bfff-ffffffffffff

When dividing requests according to the principle described above, requests to services will provide a uniform load.

Examples

Example 1 Using two instance of Luna-Streams-Retranslator on host_1 and host_2 on default port 5250. Retransmission requests will disturbed between services on host_1 and host_2, all other requests (/version, /1/docs/spec, etc) will sent to service on host_1.

Nginx config example

 http {
     server {
         listen 5250;
         location ~ /1/streams/[0-7].* {
             proxy_pass http://host_1:5250;
         }
         location ~ /1/streams/[8-9|a-f].* {
             proxy_pass http://host_2:5250;
         }
         location / {
             proxy_pass http://host_1:5250;
         }
     }
     server {
         listen 8888;
         location ~ /[0-7].* {
             proxy_pass http://host_1:8888;
         }
         location ~ /[8-9|a-f].* {
             proxy_pass http://host_2:8888;
         }
     }
 }

Example 2 Using four instance of Luna-Streams-Retranslator on host_1, host_2, host_3, host_4 on default port 5250. Retransmission requests will disturbed between services on host_1, host_2, host_3, host_4, all other requests (/version, /1/docs/spec, etc) will sent to service on host_1.

Nginx config example

Nginx config example

 http {
     server {
         listen 5250;
         location ~ /1/streams/[0-3].* {
             proxy_pass http://host_1:5250;
         }
         location ~ /1/streams/[4-7].* {
             proxy_pass http://host_2:5250;
         }
         location ~ /1/streams/[8-9|a-b].* {
             proxy_pass http://host_3:5250;
         }
         location ~ /1/streams/[c-f].* {
             proxy_pass http://host_4:5250;
         }
         location / {
             proxy_pass http://host_1:5250;
         }
     }
     server {
         listen 8888;
         location ~ /1/streams/[0-3].* {
             proxy_pass http://host_1:8888;
         }
         location ~ /1/streams/[4-7].* {
             proxy_pass http://host_2:8888;
         }
         location ~ /1/streams/[8-9|a-b].* {
             proxy_pass http://host_3:8888;
         }
         location ~ /1/streams/[c-f].* {
             proxy_pass http://host_4:8888;
         }
     }
 }