...
On a particular node N, peer nodes are stored in a table T. There 256 (ish) rows in the table, with each row's entries being of a distance from N under the above XOR metric. Row size is capped by a global parameter k, called redundancy. There is some implementation wiggle room in policies for node inclusion and eviction, which govern how the redundancy is maintained.
| Info |
|---|
| title | Who's stored in the routing table |
|---|
|
|
...
...
...
...
...
| ebd029db99cc | | pageCount | 1 |
|---|
| instanceId | 7bf7a548-f937-390d-b7ee-64c4940fe12e |
|---|
| pages | |
|---|
| width | 700 |
|---|
| documentId | 0b921171-bcf9-40d1-8af8-3cb203975791 |
|---|
| align |
|---|
|
|
...
...
...
...
Throughout the life of N, qualitative data about peers or N's link to its peers in T may be updated. For instance, if we're interested in latency-based routing, that information can be accumulated in the table. It's possible that T becomes useful beyond node routing, in which case even reputation data could be stored there.
...