Skip to content

Latest commit

 

History

History
267 lines (177 loc) · 12.7 KB

beacon-network.md

File metadata and controls

267 lines (177 loc) · 12.7 KB
Raw

Beacon Chain Network

Notice: This document is a work-in-progress for researchers and implementers.

This document is the specification for the Portal Network overlay network that supports the on-demand availability of Beacon Chain data.

Overview

A beacon chain light client could keep track of the chain of beacon block headers by performing Light client state updates following the light client sync protocol. The LightClientBootstrap structure allow setting up a LightClientStore with the initial sync committee and block header from a user-configured trusted block root.

Once the client establishes a recent header, it could sync to other headers by processing objects of type LightClientUpdate, LightClientFinalityUpdate and LightClientOptimisticUpdate. These data types allow a client to stay up-to-date with the beacon chain.

To verify canonicalness of an execution block header older than ~27 hours, we need the ongoing BeaconState accumulator (state.historical_summaries) which stores Merkle roots of recent history logs.

The Beacon Chain network is a Kademlia DHT that forms an overlay network on top of the Discovery v5 network. The term overlay network means that the beacon chain network operates with its routing table independent of the base Discovery v5 routing table and uses the extensible TALKREQ and TALKRESP messages from the base Discovery v5 protocol for communication.

The TALKREQ and TALKRESP protocol messages are application-level messages whose contents are specific to the Beacon Chain Light Client network. We specify these messages below.

The Beacon Chain network uses a modified version of the routing table structure from the Discovery v5 network and the lookup algorithm from section 2.3 of the Kademlia paper.

Data

Types

  • LightClientBootstrap
  • LightClientUpdate
  • LightClientFinalityUpdate
  • LightClientOptimisticUpdate
  • HistoricalSummaries

Light client data types are specified in light client sync protocol.

Retrieval

The network supports the following mechanisms for data retrieval:

  • LightClientBootstrap structure by a post-Altair beacon block root.
  • LightClientUpdatesByRange - requests the LightClientUpdate instances in the sync committee period range [start_period, start_period + count), leading up to the current head sync committee period as selected by fork choice.
  • The latest LightClientFinalityUpdate known by a peer.
  • The latest LightClientOptimisticUpdate known by a peer.
  • The latest HistoricalSummaries known by a peer.

Specification

Distance Function

The beacon chain network uses the stock XOR distance metric defined in the portal wire protocol specification.

Content ID Derivation Function

The beacon chain network uses the SHA256 Content ID derivation function from the portal wire protocol specification.

Wire Protocol

The Portal wire protocol is used as the wire protocol for the Beacon Chain Light Client network.

Protocol Identifier

As specified in the Protocol identifiers section of the Portal wire protocol, the protocol field in the TALKREQ message MUST contain the value of 0x501A.

Supported Messages Types

The beacon chain network supports the following protocol messages:

  • Ping - Pong
  • Find Nodes - Nodes
  • Find Content - Found Content
  • Offer - Accept

Ping.custom_data & Pong.custom_data

In the beacon chain network the custom_payload field of the Ping and Pong messages. The first packet between another client MUST be Type 0: Client Info, Radius, and Capabilities Payload. Then upgraded to the latest payload supported by both of the clients.

List of currently supported payloads, by latest to oldest.

Routing Table

The Beacon Chain Network uses the standard routing table structure from the Portal Wire Protocol.

Node State

Data Storage and Retrieval

Nodes running the beacon chain network MUST store and provide all beacon light client content for the range as is specified by the consensus light client specifications: https://github.com/ethereum/consensus-specs/blob/dev/specs/altair/light-client/full-node.md#deriving-light-client-data

This means that data radius and the concept of closeness to data is not applicable for this content.

When a node cannot fulfill a request for any of this data it SHOULD return an empty list of ENRs. It MAY return a list of ENRs of nodes that have provided this data in the past.

When a node gossips any of this data, it MUST use random gossip instead of neighborhood gossip.

Data Radius

The Beacon Chain Network includes one additional piece of node state that should be tracked. Nodes must track the data_radius from the Ping and Pong messages for other nodes in the network. This value is a 256 bit integer and represents the data that a node is "interested" in.

We define the following function to determine whether node in the network should be interested in a piece of content:

interested(node, content) = distance(node.id, content.id) <= node.radius

A node is expected to maintain radius information for each node in its local node table. A node's radius value may fluctuate as the contents of its local key-value store change.

A node should track their own radius value and provide this value in all Ping or Pong messages it sends to other nodes.

Data Types

The beacon chain DHT stores the following data items:

  • LightClientBootstrap
  • LightClientUpdate

The following data objects are ephemeral and we store only the latest values:

  • LightClientFinalityUpdate
  • LightClientOptimisticUpdate
  • HistoricalSummaries

Constants

We use the following constants from the beacon chain specs which are used in the various data type definitions:

# Maximum number of `LightClientUpdate` instances in a single request
# Defined in https://github.com/ethereum/consensus-specs/blob/dev/specs/altair/light-client/p2p-interface.md#configuration
MAX_REQUEST_LIGHT_CLIENT_UPDATES = 2**7  # = 128

# Maximum number of `HistoricalSummary` records
# Defined in https://github.com/ethereum/consensus-specs/blob/dev/specs/phase0/beacon-chain.md#state-list-lengths
HISTORICAL_ROOTS_LIMIT = 2**24  # = 16,777,216

ForkDigest

4-byte fork digest for the current beacon chain version and genesis_validators_root.

LightClientBootstrap

light_client_bootstrap_key = Container(block_hash: Bytes32)
selector                   = 0x10

content                    = ForkDigest + SSZ.serialize(LightlientBootstrap)
content_key                = selector + SSZ.serialize(light_client_bootstrap_key)

LightClientUpdatesByRange

light_client_update_keys   = Container(start_period: uint64, count: uint64)
selector                   = 0x11

content                    = List(ForkDigest + LightClientUpdate, limit=MAX_REQUEST_LIGHT_CLIENT_UPDATES)
content_key                = selector + SSZ.serialize(light_client_update_keys)

If a node cannot provide one of the LightClientUpdate objects in the the requested range it MUST NOT reply any content.

LightClientFinalityUpdate

light_client_finality_update_key  = Container(finalized_slot: uint64)
selector                          = 0x12

content                           = ForkDigest + SSZ.serialize(light_client_finality_update)
content_key                       = selector + SSZ.serialize(light_client_finality_update_key)

The LightClientFinalityUpdate objects are ephemeral and only the latest is of use to the node. The content key requires the finalized_slot to be provided so that this object can be more efficiently gossiped. Nodes should decide to reject an LightClientFinalityUpdate in case it is not newer than the one they already have. For FindContent requests, a node will either know the last previous finalized slot, if it has been following the updates, or it will have to guess slots that are potentially finalized.

LightClientOptimisticUpdate

light_client_optimistic_update_key   = Container(optimistic_slot: uint64)
selector                             = 0x13

content                              = ForkDigest + SSZ.serialize(light_client_optimistic_update)
content_key                          = selector + SSZ.serialize(light_client_optimistic_update_key)

The LightClientOptimisticUpdate objects are ephemeral and only the latest is of use to the node. The content key requires the optimistic_slot (corresponding to the signature_slot in the the update) to be provided so that this object can be more efficiently gossiped. Nodes should decide to reject an LightClientOptimisticUpdate in case it is not newer than the one they already have. For FindContent requests, a node should compute the current slot based on its local clock and then use that slot as a starting point for retrieving the most recent update.

HistoricalSummaries

Latest HistoricalSummariesWithProof object is stored in the network every epoch, even though the historical_summaries only updates every period (8192 slots). This is done to have an up to date proof every epoch, which makes it easier to verify the historical_summaries when starting the beacon light client sync.

HistoricalSummariesProof = Vector[Bytes32, 5]

historical_summaries_with_proof = HistoricalSummariesWithProof(
    epoch: uint64,
    # HistoricalSummary object is defined in consensus specs:
    # https://github.com/ethereum/consensus-specs/blob/dev/specs/capella/beacon-chain.md#historicalsummary.
    historical_summaries: List(HistoricalSummary, limit=HISTORICAL_ROOTS_LIMIT),
    proof: HistoricalSummariesProof
)

historical_summaries_key   = Container(epoch: uint64)
selector                   = 0x14

content                    = ForkDigest + SSZ.serialize(historical_summaries_with_proof)
content_key                = selector + SSZ.serialize(historical_summaries_key)

A node SHOULD return the latest HistoricalSummariesWithProof object it has in response to a FindContent request. If a node cannot provide the requested or newer HistoricalSummariesWithProof object, it MUST NOT reply with any content.

Algorithms

Random Gossip

We use the term random gossip to refer to the process through which content is disseminated to a random set DHT nodes.

The process works as follows:

  • A DHT node is offered piece of content that is specified to be gossiped via random gossip.
  • The node selects a random node from a random bucket and does this for n nodes.
  • The node offers the content to the n selected nodes.

Validation

LightClientBootstrap

While still light client syncing a node SHOULD only allow to store an offered LightClientBootstrap that it knows to be canonical. That is, a bootstrap which it can verify as it maps to a known trusted-block-root. E.g. trusted-block-root(s) provided through client config or pre-loaded in the client.

Once a node is light client synced, it can verify a new LightClientBootstrap and then store and re-gossip it on successful verification.

LightClientUpdate

While still light client syncing a node SHOULD NOT store any offered LightClientUpdate. It SHOULD retrieve the updates required to sync and store those when verified.

Once a node is light client synced, it can verify a new LightClientUpdate and then store and re-gossip it on successful verification.

LightClientFinalityUpdate & LightClientOptimisticUpdate

Validating LightClientFinalityUpdate and LightClientOptimisticUpdate follows the gossip domain(gossipsub) consensus specs.