2.4 KiB
| title | name | category | tags | editor | contributors | |
|---|---|---|---|---|---|---|
| REQ-RES-RELIABILITY | Request-response protocols reliability | Best Current Practice |
|
Oleksandr Kozlov <oleksandr@status.im> |
Abstract
This RFC describes set of instructions used across different WAKU2 implementations for improved reliability in request-response protocols such as WAKU2-LIGHTPUSH, WAKU2-FILTER and WAKU2-STORE.
Motivation
Descriptions of mentioned protocols do not define some of the real world use cases that are oftenly observed in unreliable network environment. Such use cases can be: recovery from offline state, decrease rate of missed messages, increase probability of messages being broadcasted.
Suggestions
Peer and connection management
Each protocols should retain a pool of reliable peers. In case a protocol failed to use any peer more than once - connection to it should be dropped and new peer should be added to the pool instead.
Light Push
In case sending message failed - node should try to re-send message after some interval.
Filter
-
To decrease chances of missing messages a node can initiate more than one subscription through Filter protocol to the same content topic and filter out duplicates. This will increase bandwidth consumption and would depend on the information exchanged under content topic in use.
-
In case a node goes offline while having an active subscription - it is important to do ping again right after node appears online. In case ping fails - re-subscribe request should be fired.
Store
- none for now
Security/Privacy Considerations
If there are none, this section MAY state that fact. This section MAY contain additional relevant information, e.g. an explanation as to why there are no security consideration for the respective document.
Copyright
Copyright and related rights waived via CC0.
References
A list of references.