What did the spam attack actually break?

I've seen some confusion around what the spam attack actually broke.
From what I've gathered, it was either (a) network saturation, or (b) nodes not being able to handle the mempool (or whatever the Nano equivalent is called) throughput (due to lower-spec CPU's or something). Is either of these correct or was it something else?

1 Like

My understanding is this, and happy to be corrected.
In summary, the network continued to confirm transactions, that is, online voting weight never dropped below the level at which the network stalls. Confirmation continued amongst higher bandwidth nodes while some lower bandwidth nodes fell behind in synchronisation and were unable to process transactions. Some of these nodes powered wallets that people use to transact which resulted in delays. As those nodes came back online with upgrades and/or patches despite ongoing spam (Natrium, WeNano etc) people were able to transact again.
Dynamic PoW did not meet its expected utility initially because of the heterogenous nature of the resources in each node that meant active difficulty initially didn't rise. Once the majority weight nodes reduced their bandwidth helping the active difficulty rise, the services utilising distributed PoW also apparently improved the way they handle dynamic PoW. This resulted in transactions from those services being appropriately prioritised.

6 Likes

Ah, thank you, that makes perfect sense.