Hey everyone!
We just put up a blog post to reopen the discussion about raising the minimum fee for Stellar-network transactions. Please read that post, think about it, and leave your feedback on this thread.
--
You received this message because you are subscribed to the Google Groups "Stellar Developers" group.
To unsubscribe from this group and stop receiving emails from it, send an email to stellar-dev...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/stellar-dev/1a7aaed9-1fe1-403c-b382-bb44a2b831a5n%40googlegroups.com.
Fully agree with the proposal to raise fees. However, 0.001 XLM is still a rather small fee for executing operations on the distributed blockchain. While we'll definitely see a decrease in failed arbitrage transactions on the ledger, it doesn't eliminate all potential problems.
To my mind, we need to set a long-term target in 0.001 USD for the base_fee parameter and increase max_tx_set_size to at least 2000 ops per ledger. And here is why:
First of all, history showed that the price of XLM is highly volatile, so it's very likely that the validators will need to adjust fees relatively often to keep up with the XLM price when it goes up or down. There is no point to start a broad discussion with the community, developers, and all interested parties each time the fee has to be re-adjusted. We can all agree that fees should be nominated in, say, USD or EUR. If the community agrees to set a 0.001 USD target for the base_fee, validators can decide to vote for the fee changes in response to market price actions and coordinate the voting without starting a long process of gathering feedback from all interested parties. Of course, this only applies if the new base_fee is close to the long-term target nominated in USD (or EUR) that was previously approved by the community.
The 0.0001 USD fee (as proposed in SDF blogpost) is too small to prevent resource exhaustion attacks on validators. Currently, the Stellar mainnet can process roughly 17,280,000 operations daily (max 1000 ops per ledger, ~12 ledgers per minute). So if someone decides to flood the network filling all ledgers with meaningless transactions, it will cost only 17,280 XLM (≈ 1,700USD at current price). At the same time, validators pay a lot more for processing and storing millions of those meaningless ops.
So I propose to set at list 10x larger fees to make such an attack much less tempting. The 0.001 USD for transferring money or trading on the DEX is still significantly lower than most other blockchains or centralized solutions can offer at the moment, but it will protect validators and allow them to plan maintenance costs in the long run. The fully synced Stellar Core server requires about 2TB of SSD disk space (it's not feasible to sync the network for a new validator if the server is running on chipper low-cost HDDs – it literally takes months to do it from scratch). We'll have to upgrade our servers in the nearest future as almost no free space left on the SSDs.
And lastly, why increase the limit to 2000 ops per ledger? Not merely to show that we can. It will double the ledger capacity making the network more resilient to sudden transaction outbursts. For example, in cases when many people (and bots) are trying to put or remove DEX offers in the reaction to the anchored tokens price swings. We've seen this on Ethereum – smart contracts simply can't submit a transaction to change the price or liquidate the margin/borrowing position when the Ethereum network is under heavy load. It's one of the biggest problems for DeFi apps. Also, we know that Stellar can handle a much larger load. The limit was reduced from 5000 to 1000 ops when validators voted for protocol v11, which changed how the max_tx_set_size logic works. To my mind, a gradual increase of the network capacity (for instance, doubling max_tx_set_size twice a year) will allow us to reach the throughput comparable to Visa or Swift – is one of the most obvious requirements for achieving ultimate Stellar goals. At the same time, it allows validators to test and adjust their hardware with each max_tx_set_size upgrade without risking the entire network downtime.
Therefore, my proposal is to use 0.001 USD as a long-term base_fee target, which means that we need to vote for setting base_fee to 0.01 XLM and increase max_tx_set_size to 2000 ops per ledger.
Hey everyone!
We just put up a blog post to reopen the discussion about raising the minimum fee for Stellar-network transactions. Please read that post, think about it, and leave your feedback on this thread.
--
You received this message because you are subscribed to the Google Groups "Stellar Developers" group.
To unsubscribe from this group and stop receiving emails from it, send an email to stellar-dev...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/stellar-dev/e445664e-422d-474d-9b77-c333fcb57d2an%40googlegroups.com.
We’re fine with 10,000 stroops at the current price of $0.10/XLM.
Even at $1.00/XLM that is still incredibly low for financial transactions, but is cost-prohibitive for “stamping” transactions, such as charging for each email, since revenue for those types of micro-transactions tops out at around $1.00 CPM (.001 USD/tx).
An interim raise to 1,000 stroops for now would give validators some relief and disincentive some pigeons, while leaving room for increased valuation of XLM and use of Stellar network for micro-transactions.
However, if the long-term focus for XLM is going to be on financial transactions, then perhaps the trajectory should be towards a .01 USD/ $10CPM - still awesomely low even for sub 1USD/EUR/100Yen transactions — but impossibly high for “stamping” transactions like email.
To view this discussion on the web visit https://groups.google.com/d/msgid/stellar-dev/36b3c352-1b9c-49ee-97bd-8a1c13562c31n%40googlegroups.com.
--
You received this message because you are subscribed to the Google Groups "Stellar Developers" group.
To unsubscribe from this group and stop receiving emails from it, send an email to stellar-dev...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/stellar-dev/56e80810-56bf-42cb-9c11-17d3468a54b2n%40googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/stellar-dev/337c2050-2545-494d-8b45-80794cec16dbo%40googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/stellar-dev/1f780542-fa02-403a-9ef9-963ad9542ed3n%40googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/stellar-dev/CABf4b3z1AWEbM49YnBzC26keMqH2u%2Bm24HMEfyKmQDLx2XGr2g%40mail.gmail.com.
--
You received this message because you are subscribed to the Google Groups "Stellar Developers" group.
To unsubscribe from this group and stop receiving emails from it, send an email to stellar-dev...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/stellar-dev/8b2f3a72-d346-4e7f-9a13-b36e0b590c2ao%40googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/stellar-dev/CAHzj05UGuvh4QnjCmK1GriMYq%2BTXS_as3a1qaZRK2cjn5N%2BYaQ%40mail.gmail.com.
It looks like this whole discussion comes down to a question about how people (or rather organizations) are using the network – a payment platform or general-purpose storage.
It's a bit weird to read that some business models are not sustainable with transaction fees as low as 0.001 XLM. In most cases, it's even cheaper than storing data on your own servers, given that you need to pay for at least two servers (a minimal fail-safe cluster), system administrators to maintain them, and developers. Furthermore, what's the plan in case of surge pricing? Temporarily turn off the service each time fees go higher due to increased on-chain activity? What if surge pricing will be in action for a few weeks? History showed that Ethereum projects that couldn't afford to pay competitive gas prices ended up in oblivion. Speaking of, people doing arbitrage trading and liquidity mining on Ethereum were ready to pay dozens of dollars in fees per transaction right now due to network congestion. So the assumption that rising fees will drive off the liquidity and market makers from Stellar DEX looks very unlikely. Especially given that after the rise, they still will be like 1000x lower than Ethereum fees.
Business models built on the assumption that the storage is free (or almost free) can't be successful in the long run. Currently, all expenses are shifted to validators – they pay for the storage. The imbalance of the validator servers maintenance cost and network fees provide an opportunity for all kinds of inept project ideas. For example, I remember someone planned to create a decentralized Twitter on Stellar, storing tweets in transaction memos. Awfully inefficient, but feasible due to negligible fees. As I see it, the base fee should be large enough to prevent devs from building Stellar apps and monetization models that rely on super-cheap transactions. This also falls in line with the primary Stellar mission – to create a universal global payment network.
We all know that the ledger size keeps growing almost exponentially. Consequently, the Horzion database grows significantly larger over time. Currently, it doesn't look like a huge problem, but let's look a bit deeper. We need to understand that in a year or two, the DB will be so large that it will require expensive hardware. The cost of storing this tx in the fully ingested database will double or even triple every year due to the rising hardware requirements. Paying 20-40K USD monthly for a Horizon server hosting in a few years might result in a more centralized network, with only a few large players that can afford it.
Even now, the Horizon server has difficulties running some queries due to the massive DB and indexes. This results in timeouts when fetching payments. And it's only the beginning. When the size of the database doubles, it may result in the exponential growth of response time on complex queries since the DB engine requires more time to process each index participating in the query, then build results intersection, and fetch the data. But here is a more subtle and somewhat counterintuitive example of sudden performance degradation. Once the most frequently used indexes no longer fit in the server RAM, the overall performance of queries that rely on those indexes may drop by two orders of magnitude as the DB engine struggles to load parts of indexes from disk, perform a scan, clean up memory, and then reload another portion of the index for the new query.
Yes, SDF and community developers are conducting research on sidechains, sharding, and other scaling approaches. Maybe this problem will be completely irrelevant next year. But the data stored on Stellar pubnet right now will be there forever even after the sharding (or whatever will be the solution) is deployed. We have technical problems with processing those large volumes of economically meaningless transactions right now. What's the point of super cheap arbitrage or data storage if retrieving historical data from the ledger becomes a problem?
Ultra-cheap Stellar transactions is not a feature; it's a delayed bomb. The only reason why the fees are so low, is that there wasn't much of an on-chain activity in the beginning. Ledgers were mostly empty, so there was no need to adjust fees. Our community was ready to increase base_fee more than two years ago (there was an extensive discussion on Slack back then), but we couldn't do this as pre-signed smart contracts would be stuck forever. Now, with CAP-15 implemented, we can finally raise the base fee amount without messing up old smart contracts.
To sum it up, increasing the base fee to 0.001 USD will ensure that projects developing on Stellar right now will start planning their on-chain activities more wisely. If we can keep the ledger relatively clean for now by enforcing a bit higher fees, there is a fair chance that maintenance expenses won't rise exponentially as they will be partially compensated by the hardware evolution – disk space and memory gets cheaper every year. The higher base_fee is not a whim – it's an investment into a sustainable future of Stellar Network. And, frankly, it's a really small price to pay considering the fee statistics of most other blockchains.
Hey everyone!
We just put up a blog post to reopen the discussion about raising the minimum fee for Stellar-network transactions. Please read that post, think about it, and leave your feedback on this thread.
--
You received this message because you are subscribed to the Google Groups "Stellar Developers" group.
To unsubscribe from this group and stop receiving emails from it, send an email to stellar-dev...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/stellar-dev/e445664e-422d-474d-9b77-c333fcb57d2an%40googlegroups.com.
A group of us on Stellar Global would increasing the minimum fee. Also, we would like to see this conversation re-opened. There seemed to be a large amount of consensus building around raising the fee, but the conversation stalled. Can we get this conversation rolling again?Thanks!