Experimental FlatTrie for archive nodes (v2.1.0) – when should you use it?

English Kaia
1

20251125_Flatten the State, Shrink the Disk (EN)
20251125_Flatten the State, Shrink the Disk (EN)1200×630 85.9 KB

We’re introducing an experimental FlatTrie mode for archive nodes in v2.1.0.

As Kaia’s state grows, archive nodes become one of the biggest cost centres for anyone running:

  • explorers

  • indexers

  • long-term analytics/monitoring infra

FlatTrie is a flatter state layout for archive nodes. The goal is to:

  • make deep historical reads more cache-friendly, and

  • uses substantially less disk space compared to the current archive scheme.

At a high level, the article covers:

  • What FlatTrie is and how it changes the archive state structure

  • Why it can reduce disk usage for archive nodes

  • How to enable it (opt-in flag, fresh sync from empty DB)

  • Important limitations (archive-only mode, some RPC/rewind features not supported)

If you are:

  • planning new archive infra on Kaia,

  • running or building explorers/indexers, or

  • deciding whether you really need archive vs non-archive nodes,

You should understand how this mode works before finalising your architecture.

Full deep dive here: https://medium.com/kaiachain/flatten-the-state-shrink-the-disk-90c3db0b5fe7

Use this thread for follow-up questions (e.g., topology, hardware assumptions, migration strategy), and we’ll route them to the relevant infra folks where needed.

2개의 좋아요
2

As part of the v2.1.0 infra series, we’ve published a deep dive on Kaia’s experimental FlatTrie mode for archive nodes.

Completed infographics (A4) (1)
Completed infographics (A4) (1)1587×2245 415 KB

The short version:
– It’s archive-only and opt-in.
– It flattens the archive state layout so historical reads hit disk in a more cache-friendly way.
– For explorer/indexer/analytics workloads, this can translate into a substantially smaller archive footprint and smoother historical queries.

Important caveats:
– You must sync from an empty data directory with the FlatTrie flag enabled; you can’t convert an existing archive DB.
– Some RPCs and state operations are not supported in this experimental mode, so it’s intended for advanced operators who can test in staging before deploying.

If you’re planning new archive infra on Kaia or revisiting how you run explorers/indexers, it’s worth reading the full article and considering FlatTrie as a specialised option in your stack.

Full write-up: https://medium.com/kaiachain/flatten-the-state-shrink-the-disk-90c3db0b5fe7

1개의 좋아요