vac:dst:waku:2025q1-waku-scaling
Description
We will use realistic Waku node testing, theoretical analysis, and simulations to evaluate and improve Waku’s scalability. Performance measurements in waku components will be also done. This includes identifying Waku’s limits and measuring performance in various scenarios, delivering concrete data and actionable insights to validate or challenge our ideas.
Background
Narratives
These efforts will support the Conduit of Expertise narrative by accelerating Waku Improvements, providing measurable insights to enhance developer experience within and beyond the IFT ecosystem. Enhancing RFC Development by enabling faster, more informed RFC development through performance insights during the drafting process.
Task list
Scale one to one chat
Note: This needs more input from project
- fully qualified name:
vac:dst:waku:2025q1-waku-scaling:scale-one-to-one - owner: Alberto
- status: 0%
- start-date: <yyyy/mm/dd>
- end-date: <yyyy/mm/dd>
Description
Deliverables
Messaging API
Note: This needs more input from project
- fully qualified name:
vac:dst:waku:2025q1-waku-scaling:messaging-api - owner: Alberto
- status: 0%
- start-date: <yyyy/mm/dd>
- end-date: <yyyy/mm/dd>
Description
- Includes compare gowaku with nwaku replacement?
Deliverables
Direct message reliability
- fully qualified name:
vac:dst:waku:2025q1-waku-scaling:direct-message-reliability - owner: Alberto
- status: 0%
- start-date: <yyyy/mm/dd>
- end-date: <yyyy/mm/dd>
Description
Store message hash queries
of less than 10 hashes
are served (wire-to-wire) under 400 ms,
as long as the database
is less than 250 GB size
and average message size under 500KiB.
This refers to message-hash-queries
that have include-data=true.
Wire-to-wire includes the whole storage time, as in database-time + transport-time + node-time.
Store message time range queries of less than 24 hours, with less than 10 content topics, are served under 400 ms, as long as the database is less than 250 GB size and average message size under 500KiB.
To validate this, it will be used against a status-prod-like database.
Deliverables
- Analysis report
- Code
Decentralized storage dogfooding
Note: This needs more input from project
- fully qualified name:
vac:dst:waku:2025q1-waku-scaling:decentralized-storage-dogfooding - owner: Alberto
- status: 0%
- start-date: <yyyy/mm/dd>
- end-date: <yyyy/mm/dd>
Description
Deliverables
Contact discovery app features
Note: This needs more input from project
- fully qualified name:
vac:dst:waku:2025q1-waku-scaling:discovery-app-features - owner: Alberto
- status: 0%
- start-date: <yyyy/mm/dd>
- end-date: <yyyy/mm/dd>
Description
Deliverables
Store sync
Note: This needs more input from project
- fully qualified name:
vac:dst:waku:2025q1-waku-scaling:store-sync - owner: Alberto
- status: 0%
- start-date: <yyyy/mm/dd>
- end-date: <yyyy/mm/dd>
Description
Analyze store sync Waku’s functionality, where store nodes get in sync. This should happen after a disconnection, or when a fresh node is joined to the network.
Deliverables
- Analysis report
- Code
Waku Shard Scaling
- fully qualified name:
vac:dst:waku:2025q1-waku-scaling:waku-shard-scaling - owner: Alberto
- status: 100%
- start-date: 2025/01/06
- end-date: 2025/01/24
Description
Address feedback from vac:dst:waku:2024q4-waku-scaling:waku-shard-reliability-vs-scale.
Increase the number of shards up to the highest possible value.
Additional info
Blocked by a nwaku issue. Another task will be created to keep requested work.
Deliverables
- Reports:
- Issue found: Issue
- Related PRs if apply:
- Time to get healthy plot PR: PR
- Blocking issue found: Github Issue
PX vs speed of connections
- fully qualified name:
vac:dst:waku:2025q1-waku-scaling:px-speed-connections - owner: Wings
- status: 0%
- start-date: 2025-02-03
- end-date: 2025-02-14
Description
Study the effect of Peer Exchange in a context where a high number of nodes are not reachable anymore.
Run a simulation with a high X (1000) number of nodes, wait until full discv5 is done, stop half of those nodes, wait for N minutes (N=0,1,5,10), start Y (100) new nodes, and measure discovery time.
Deliverables
- Reports:
- Related PRs if apply:
Discv5 initial activity
- fully qualified name:
vac:dst:waku:2025q1-waku-scaling:discv5-initial-activity - owner: Wings
- status: 0%
- start-date: 2025-02-10
- end-date: 2025-02-21
Description
Ensure data before stable mesh do not skew results. Also, analyze Discv5 usage in this scenario. Create a simulation without traffic, where nodes are leaving and joining the network, once the mesh is healthy. Compare these results from initial state and healthy state. Try this with different network sizes. Help to identify scenarios that are most realistic.
Deliverables
- Reports:
- Related PRs if apply: