Tari Protocol Discussion 45
On Monday, the Tari community discussed multiparty payment channels
Join us for our next discussion on Freenode in #tari-dev.
Discussion times proposed by the Tari community:
Mondays: 6pm CAT (12pm EST)
Thursdays: 11am CAT (5am EST)
To keep up with the latest Tari developments, you can follow the project on Twitter.
Transcript of Monday discussion
18:03 <@CjS77> Evening, everyone 18:03 <Hansie> W.r.t. scaling, I am not overly optimistic that the MW base layer can sustain a vibrant DAN. 18:04 <@CjS77> You're being a bit kind :) I'd say it's impossible 18:04 <Hansie> Current throughput in practice of MW is around 17 tps 18:05 <Hansie> cjs77: Agreed 18:05 <@CjS77> You simply can't map the type of throughput a digital asset like a set of tickets for a popular event going on sale to a PoW blockchain 18:05 <@CjS77> I think this is discussed in one of the RFCs.. 18:06 <Hansie> So how do we scale? 18:06 <@CjS77> https://rfc.tari.com/RFC-0001_overview.html 18:07 <@CjS77> The RFC mentions payment channels as a means of bridging payments on the DAN to the base layer 18:08 <@CjS77> Obv Lightning is the most well-known payment channel solution 18:08 <Hansie> Something like Lightning perhaps? 18:08 <@CjS77> Yeah 18:09 <@CjS77> But I think we can take advantage of Mimblewimble's properties to come up with something a little more ergonomic. 18:09 <Hansie> Agreed, Lightning to my mind is somewhat complicated and restricted for the DAN 18:09 <@CjS77> Possibly with a different set of guarantees, but one that is still fast, secure and essentially trustless 18:10 <Blackwolfsa> the only problem with lighting is the that it can be seen as many strings with beads on them, not very flexable.. 18:10 <@CjS77> Yeah, only allowing bilateral channels is something of a deathknell for non-fungiblr token 18:10 <Blackwolfsa> and its only two parties 18:10 <Hansie> The lack of scripting system in MW could be a potential drawback 18:10 <@CjS77> And the routing is very complicated 18:11 <@CjS77> Hansie - yes it makes things trickier.. 18:11 <Blackwolfsa> technically its very possible to do two party payment channels in mw 18:12 <@CjS77> Indeed, are you referring to Beam's recent release? 18:13 <Hansie> Beam has a demo at this stage, with plans to take it to testnet then mainnet 18:13 <@CjS77> For Tari, I think it would be preferable for the payment channel topology to map the validator node topology very closelt 18:14 <@CjS77> i.e. multiparty payment channels, with parties being able to join (deposit) and leave (withdraw) roughly at will 18:14 <Hansie> Do you mean permissionless? 18:14 <Hansie> Ah ,ok 18:14 <Blackwolfsa> not even, the math/tech/know how to do it there 18:14 <Blackwolfsa> but I dont think its really the best of ideas 18:15 <Blackwolfsa> having a "built in" multiparty payment channel is a much better idea I think 18:15 <Hansie> "built-in" or "bolt-on"? 18:16 <@CjS77> What do you mean by built-in? 18:19 <Blackwolfsa> natively supported 18:19 <Blackwolfsa> its not just some extra after though or hacked on feature 18:20 <Hansie> Ah, do you mean with all the proper hooks? 18:20 <@CjS77> I see. yeah, although that said, the less you have to add to standard MW consensus rules, the better 18:21 <Blackwolfsa> true.. 18:21 <Hansie> Lobbyists have been advocating for a long time now to add hooks to Bitcoin in support of side channels 18:21 <@CjS77> lol. "Lobbyists" 18:22 <Hansie> Ok, enthusiasts then? 18:22 <@CjS77> I'd love to see Paul Storcz's reaction at being called a lobbyist :) 18:23 <Hansie> oops 18:23 <@CjS77> Too funny 18:23 <Hansie> I can only claim to be native Afrikaans 18:23 <@CjS77> sarang, have you given any thought to this at all? We have some ideas that we can bounce off you 18:25 <@CjS77> We can expand on this next week perhaps. Seems you're not around today. 18:27 <Hansie> Yip, maybe enough food for thought for now? 18:28 <@CjS77> 👍 18:50 <sarang> I'm here, for a bit 18:51 <sarang> We've looked into sidechannels and PCNs for Monero... there's a preprint out with an idea for how to do it with a modification to our tx protocol, but it has big drawbacks and doesn't have an obvious crossover to MW 18:51 <sarang> But yes, it's the lack of scripting that gets ya 2019-08-12 18:03 <Hansie> Hi there, time for another dev chat. If anyone is present, I thought we could continue where we left of last week, i.e. further discussing permissionless multiparty payment channels as scaling solution. 18:13 <Hansie> If we could have any number of these payment channels, managed by committees that must some consensus voting mechanism for those transactions, what would the scaling issues be? 18:17 <neonknight64> When these committees want to submit an accumulated transaction to the base layer such as a checkpoint transaction. The size of the transaction will be limited by the block size. 18:17 <Hansie> Let us assume we can have a small footprint (i.e. commitment) of the state of the payment channel represented in the base layer at certain checkpoint intervals, and that we could have a 1-1 peg. 18:18 <Hansie> neonknight64: Please explain what you mean by `accumulated transaction` 18:20 <Hansie> And yes, block size is certainly a limitation 18:20 <neonknight64> Potentially, a single large transaction that spends to large number of outputs. 18:22 <Hansie> Ah ok, so large numbers of outputs contain an equal amount of Bulletproof range proofs, thus the size limitation. 18:24 <Hansie> If I am not mistaken, a quick calculation reveals ~2,000 new UTXOs created per block 18:24 <Hansie> That is 17 tps, assume 2 outputs per transaction, x 60s. 18:25 <Hansie> With ~1,000 kernels 18:25 <Hansie> ^ as upper limit 18:28 <simian_za> So what would make up all these transaction outputs in a payment channel checkpoint? 18:28 <simian_za> individual withdrawals? 18:29 <Hansie> We could have deposits, withdrawals and possibly refunds. 18:35 <neonknight64> A committee with a large user base might have to split these base layer transaction into separate transactions to ensure they fit into blocks 18:38 <Hansie> Ok, so is it practical to assume that whatever scheme employed it will be limited by a collective creation of ~2,000 new UTXOs and ~1,000 transactions per block? As a sustained throughput. 18:40 <simian_za> Sure, so a Channel will be have to stagger/cascade a large amount of withdrawals, deposits or refunds across multiple blocks 18:41 <Hansie> Great. If the network is not busy a burst of activity could be staggered and after a while all will be dealt with. 18:43 <stanimal> Agree, greater throughput could potentially be achieved by splitting up transactions that would exceed these limits across more than one block - seems like a nice solution 18:45 <Hansie> So we have discussed block size and/or tps as an upper limit for payment channels' scaling issues. This can be used to calculate a theoretical sustained maximum throughput for all payment channels combined. 18:46 <simian_za> Well not exactly, we have discussed the throughput of deposits, withdrawals and refunds 18:47 <simian_za> if the transactions stay in the payment channel thats a different story? 18:50 <Hansie> I was trying to deduct scaling issues for payment channels from this discussion. There is a definite limit, depending on its design, if checkpoints involve multiple UTXOs from multiple users. 19:01 <Hansie> Thanks guys, this is then the end of the regular chat.