K-Bucket 在 Kademlia DHT 中究竟意味着什么？

Question

I want to confirm my understanding of buckets in Kademlia DHT. Kademlia has m k-buckets where m is the size of the network in bits and k is the number of key-value pairs stored per bucket. for example, let's say m=4 then we can have 2^4 nodes, namely from 0 to 15.

+========+
| NodeId |
+========+
|   0000 |
+--------+
|   0001 |
+--------+
|   0010 |
+--------+
|   0011 |
+--------+
|   0100 |
+--------+
|   0101 |
+--------+
|   0110 |
+--------+
|   0111 |
+--------+
|   1000 |
+--------+
|   1001 |
+--------+
|   1010 |
+--------+
|   1011 |
+--------+
|   1100 |
+--------+
|   1101 |
+--------+
|   1110 |
+--------+
|   1111 |
+--------+

Each node has the routing table of the 0-bit match, 1st-bit match and 2nd-bit match and so on, this is m buckets. Furthermore, for each bucket, it will store k representatives instead of a single NodeId. So, if we say k=2, the routing Table for node 0101 would be something like:

┌──────────────────────┐
│         0101         │
├──────────────────────┤
|                      |
| +==================+ |
| |       xxxx       | |
| +==================+ |
| |   1001, <value>  | |
| +------------------+ |
| |   1010, <value>  | |
| +------------------+ |
|                      |
| +==================+ |
| |       0xxx       | |
| +==================+ |
| |   0000, <value>  | |
| +------------------+ |
| |   0111, <value>  | |
| +------------------+ |
|                      |
| +==================+ |
| |       01xx       | |
| +==================+ |
| |   0110, <value>  | |
| +------------------+ |
| |   0111, <value>  | |
| +------------------+ |
|          .           |
|          .           |
|          .           |
└──────────────────────┘

Is my assumption correct?

Answer 1

k is the number of entries in a bucket. Their node IDs are expected to be randomly distributed within the ID-range the bucket covers, which means doubling the number of entries per bucket would only increase its resolution by one bit on average, ie it does not scale well. That's why we have a structured routing table with multiple buckets with different scope each. It increases local resolution around the node's own node ID.

Implementing the full kademlia algorithm requires a dynamic routing table layout . Therefore m is not fixed. The fixed size layout was only used in the simplified pre-print version of the paper as part of a theoretical proof.

Answer 2

Every node in Kademlia store a list of other nodes. This list is based on bit-differences and is loosely termed as buckets. Now, 'k' is a system wide replication parameter. This means that for every list, there are k entries inside for that 'bucket'. This is my understanding. This is the link to the paper. Hope this helps..:)