DittoExperimental

Undocumented

Set a Query Overlap Group for the local Ditto instance. Query Overlap Groups are an optional feature that can be used to group peers based on their anticipated queries, and set relative connection priorities amongst these groups.

The numeric value for a query overlap group is entirely use-case specific and apps are free to use whichever values they like - provided all peers agree on the meaning of the values. All peers belong to query overlap group 0 initially and will connect with normal priority until specified otherwise. The number that represents the query group doesn’t mean that the group has higher priority. The number is just an enum to represent the group. To assign priority, see setPriority().

Why Are Query Overlap Groups Needed?

Ditto LiveQueries will replicate only documents that match a query between peers. In some cases, two peers with the same query might be separated by others who do not share the same query. Consider a simple example of three peers connected in a line (A - B - C). If A and C share the same query but B does not, data will not replicate from A to C as B is not interested in the data - and therefore doesn’t possess it to replicate onwards.

If your app’s use-case includes non-overlapping queries like this example, query overlap groups are designed to help. You can use Query Overlap Groups to tell Ditto that peers A and C are likely to have queries that overlap. This extra information enables Ditto to take measures to ensure realtime sync occurs, even creating an encrypted tunnel for A - C via B. Not all apps will require this functionality. In many apps, all peers share much the same queries. These apps are better served by the automatic mesh that Ditto will create for you. If used, Query Overlap Groups should be set before starting Ditto sync.

Use Cases

High Priority Between Same Group Members

Ensuring that peers with the same query can fully replicate with each other, even if they are separated by other peers that don’t have the same queries.

For example, perhaps your use case has administrative apps which should replicate all data amongst each other, and regular apps which query only a subset of the data. You can assign all administrative apps to one query overlap group (e.g. 2). You can then configure peers with the query overlap group “2” to treat other peers in group “2” as a high-priority connection target. For all peers in a high-priority Query Overlap Group, Ditto will prioritize these connections, and set up end-to-end encrypted multihop connections if direct connections aren’t available. In doing so, our example administrative apps can be sure that their data will remain in sync, even if they’re separated by other regular apps in the local mesh.

High Priority Between Different Group Members

Ensure replication between peers in one Query Overlap Group with peers in a different Query Overlap Group - perhaps due to the role or function which those peers perform in the mesh.

For example, consider a fast-food ordering app in which customers query their order (and only their own) working together with staff who must receive everyone’s orders. We want to express that customers should replicate with the staff with high priority, whereas there’s little to be gained from customers replicating amongst each other since each of their orders are private. All customer apps can be assigned to one Query Overlap Group (e.g. 1) and all staff to another (e.g. 2). Customer apps would configure a high-priority connection preference for staff. In doing so, customer apps would establish a direct connection to staff if possible, and otherwise establish an end-to-end encrypted connection hopping via other customers if needed. The end-to-end encrypted connection guarantees customer-to-staff replication without personal information being readable by other customers along the way.

Low Priority Between Different Group Members

Ensure connections aren’t established with certain Query Overlap Groups if the connection would be wasteful.

Query Overlap Groups can also be used to prevent peers from connecting if you know the connection would be wasteful. Perhaps you know that different types of peers or apps don’t share any data in common. In this case, peers can be assigned to the relevant Query Overlap Group, and the priority can be set to the lowest value of “don’t connect”. The peers will then save their available connection capacity for other, potentially more useful connections.

Configure the priority with which Ditto should connect to any peers in the given Query Overlap Group. If the priority is set to high, an end-to-end encrypted multihop connection will be established wherever it is not possible to establish a direct connection.

Connections can be established in either direction, so the remote peers might attempt connections to this Ditto with different priorities unless they are similarly configured.