Test backing group sizes of 3 on Kusama

[eskimor]

Less backers, means more resources for approval checking. 3 should be enough in general, especially once we have asynchronous backing. Without it could have negative impacts on parachain liveness, but let's test on Kusama.

• Propose backing group size 3 on Kusama
• Once enacted, monitor parachain block times and in general network performance.
• Things became better/stayed the same -> Go for Polkadot
• Things got worse -> role back -> Create follow up ticket to re-evaluate with async backing.

[eskimor]

If the outcome is negative we should retry once async backing is fully launched (including most parachains taking advantage of it).

[ordian]

• public referendum: https://kusama.polkassembly.io/referenda/294
• fellowship whitelist: https://kusama.polkassembly.io/member-referenda/50

[ordian]

The change has been enacted in block 20319170 (you can check this e.g. by querying the activeConfig with that hash and its parent), at 2023-10-29 06:27:54 (+UTC). Taking a look at the introspector dashboard, we can see that even though there are visible spikes on parachain block times happening after the backing group size reduction, it has only slight effect on average.
We can also see it has impacted some parachains more than other probably due to collator location relative to the validators.

Some screenshots:

[eskimor]

Thanks @ordian ! Great work! Ok, I would say definitely not go for Polkadot. Roll back might not be necessary, given that async backing will be launching soon anyway. Have there been any complaints yet?

[mrcnski]

• public referendum: kusama.polkassembly.io/referenda/294

Members spoke highly of their experiences with the proposer

Nice!