Node diversity Workshop – Porto

02^nd June 2026

First of all

2

Why node diversity ? Why Amaru is leading ?

3

Many teams, many nodes, many ideas, one Cardano

We planned it in our budget proposal and we think it’s important

We need to put on paper the hard topics and move together in sync

What do we want to get out of this workshop?

4

Who are you? What do you do for Cardano ? (>20 sec)

What’s 1 pragmatic thing you want solved out of these 2 days ?

Day 2 is all about getting what you want, but people need to agree

5

Whiteboard will be the voting system: everything is free to for the taking

This afternoon at 14:00 we vote for the topics on Day 2 morning & afternoon

Tomorrow we vote again in the morning at 09:30 and at 14:00 to refine the ideas

Each sessions will have to be turned into (at least) one pragmatic action related to the topic

Now let’s do some Leios

6

Things not to break with Leios

Ryan Williams

Notes on topic discussed:

8

Things not to break with Leios

block changes

• header - new fields
• body - new fields

​

transaction in EB transaction shape changes

• new certificates

​

mini protocols

• handshake – unchanged
• Local chainsync – ongoing discussions
• chainSync - unchanged structure, just use new types blockFetch - unchanged structure, just use new types txSubmission - unchanged atm, V2 in progress
• keepAlive - unchanged
• peerSharing - unchanged
• LeiosNotify (new, needed for cosensus participation) LeiosFetch (new, needed for following change)
• LeiosVotes (potentially new)

core tools

• Indexers -- mini-protocols, block shape changes, transaction shape changes
• Serialisation Libraries -- block shape changes, transaction shape changes
• SPO tooling -- new certificate types, new keys

Downstream

• Block explorers
• Wallets
• DApps

​

Impact, high / low

impact analysis

notes from Ryan's note

Leios ecosystem impact analysis, is really good

https://github.com/input-output-hk/ouroboros-leios/blob/4603cfd0b545cccf3d7c8fddc75e6e0f182f132a/docs/ImpactAnalysis.md

​

a large consideration for tooling/DApps is keeping up with the chain -- throughput

there is a cost to make the changes for new data structures etc.

there are on-going costs for bigger infra for more blocks and transactions

(there will a phased rollout for throughput, to allow scale up of infrastructure)

(throughput in existing impact analysis doc)

some tools are built with expectation of higher throughout some are not

​

Q: Dijkstra and Leios upgrade at the same time?

one painful change vs multiple painful change for ecosystem

​

van Rossem comms

focus on the shiny stuff

rather than the breaking changes

it should of told people "you need to respond to X should be the priority"

​

localchain sync -- ongoing discussion on how to handle txs from EBs

Notes on topic discussed:

9

Things not to break with Leios

Leios ecosystem impact analysis, is really good

https://github.com/input-output-hk/ouroboros-leios/blob/4603cfd0b545cccf3d7c8fddc75e6e0f182f132a/docs/ImpactAnalysis.md

​

a large consideration for tooling/DApps is keeping up with the chain -- throughput

• there is a cost to make the changes for new data structures etc.
• there are on-going costs for bigger infra for more blocks and transactions
• (there will a phased rollout for throughput, to allow scale up of infrastructure)
• (throughput in existing impact analysis doc)
• some tools are built with expectation of higher throughout some are not

​

Q: Dijkstra and Leios upgrade at the same time?

• one painful change vs multiple painful change for ecosystem

​

van Rossem communications

• focus on the shiny stuff
• rather than the breaking changes
• it should of told people "you need to respond to X should be the priority"

​

localchain sync -- ongoing discussion on how to handle txs from EBs

Topic discussed

10

How much resources do we need to allocate to run Leios?

Assumptions:

Setup (stake distribution):

25% US

47% Europe

​

Worst case: they buy stakes and setup a local setup near the stake and they create a CERT as soon as possible, put it on chain, but they get the CERT and not the data

​

🡺 Praos say it’s the best

but Leios doesn’t allow it

​

Praos: select the longest chain; if it has a CERT it is the best chain

Linear Leios prevents you from selecting until you get all the EB related to the CERT

Node diversity leaderboard

Damien Czapla

Node diversity Leaderboard – Basic concept

12

Basic concept: a CI executes a set of tests cases to measure “how good is your Cardano node”

CI contains for each category:

• A workflow detailing the content going through the pipeline
• Expected results for each test
• Completeness analysis of the test against the expected results

Node diversity – Categories of the “Cardano node leaderboard”

13

Block propagation performance

Resources usage for “basic node activities”

Protocol level Benchmarks

Conformance

Protocol level tests

Subsystem level tests

Subsystem level benchmarks

Ledger

Consensus

Types of transaction

Size of transaction & Content

Validation & forging of a block

Transaction submission

Chain query

Adversarial scenarios compliance

Plutus VM

​

Ouroboros

Praos

Leios

Known attack vectors

Antithesis testing

White hacking tests

Expected thresholds > Pass or fail

Magnitude of performance

Expected results > Pass or fail

Magnitude of performance

Compiling

Ledger state

Node diversity – Categories of the “Cardano node leaderboard”

14

Protocol level Benchmarks

Conformance

Subsystem level benchmarks

Adversarial scenarios compliance

Global leaderboard & Local leaderboard: evaluating the performance and compliance of the nodes

Cardano node diversity workshop

03^rd June 2026

What will happen next

16

Damien will encapsulate the notes, the learning and follow-up on the pragmatic steps

We will meet again for synchronising, keep an eye in your emails and the node diversity Discord

17

Thank you

for being involved during the 2 days