Hedera Hashgraph Part 4: Does Hedera really do 10,000 TPS?
DLT: Hedera Hashgraph
12th August 2024.
Spoiler Alert: Yes it does. Or it depends.
I want to comment on this article from cointelegraph that claims that Hederas 10,000 TPS is overstated. They quote a source from a guy called Eric Wall who wrote a critical article about Hedera in 2019. To be fair, that was the medieval age of DLT when considering how fast this technology develops. Also, Hedera was just released back then so there were legitimate reasons to be skeptical.
I decided to write about it to explain why Eric’s post doesn’t hold up today. Why even bother? It’s the only Hedera related post on Coin Telegraph I can find, or at least the one that shows up first. Critics of Hedera might still use this article for their arguments.
Note: Eric Wall will be referenced as “the author” or “he/him” from now on.
Comparison with Bitcoin
I just have to comment on his claim about miners not being the issue why bitcoin is slow. You can read it yourself but I’m paraphrasing here:
My understanding is that he says it’s because the nodes are rate-limited. You could in theory increase the rate of transactions, like bitcoin cash have, but you would still not be near the 10,000 TPS of Hedera. And the side-effect would be the risk of increased forking, which could destroy the whole network. It is isn’t even possible to tweak the rate on a node without creating an entire new blockchain, because it wouldn’t be compatible with the rest of the network. All of this is stated by the author himself, so I don’t get why he is trying to defend bitcoin.
Whatever the reason is, the fact still remains in 2024: If you search on the Internet for “how many transactions per second can bitcoin do” you still get the answer that bitcoin can only do 5-10 TPS.
Now if you consider the Lightning Network as he mentions, it’s a whole other story, The LN claims to be able to, in theory, handle a million TPS. That is quite impressive, but is it as secure as Hedera? I’m no expert but It doesn’t seem to be Asynchronous Byzantine Fault Tolerant to me.
Even if we assume that LN is safe to use, a million TPS is just overkill. Mastercard does 5,000 TPS and payments rarely take longer than a few seconds to complete. Without throttling, Hederas theoretical limit is somewhere around 100,000 TPS. With sharding it could probably go up to a million as well. And Hedera is a layer 1 blockchain while LN is a layer 2 blockchain.
And he goes on about how small the bitcoin packets are and how the latency is comparable with Hedera, but none of that has anything to do with throughput. It doesn’t matter if your packet is 1 Byte in size and can be transferred in 1ms, if your throughput is 10min between every packet.
But sure, for the arguments sake let’s say bitcoin packets are smaller than Hedera. According to the author, Bitcoin packet size is 80+6 Bytes in size and Hedera is 64x10 Bytes (his assumption that it needs ~10 transactions to reach full consensus). Bytes! If the Internet connection is 1Gb/s, as the minimum requirements for Hedera validator nodes states, 640 Bytes represents 0.000512% of the total capacity. I doubt that the extra overhead of Hedera will saturate a validators internet connection any time soon.
Comparison with Ethereum
He then goes on to compare Hedera with Ethereum and that is where we find the core argument about Hederas 10,000 TPS being overstated. I quote him here:
“The only thing the hashgraph does is sort the transactions to be received by the EVM [Ethereum Virtual Machine]. But the problem for the EVM isn’t that it can’t receive 10,000 ordered TPS, it’s that it can’t process the transactions quickly enough even on a fast SSD. You’d only make the queue longer.”
The Ethereum Virtual Machine is used to handle smart contracts on the Ethereum Network. The Ethereum Network’s throughput is, according to the most recent article I can find, around 27 TPS. The TPS Dashboard however constantly shows 14.4 TPS.
Note: Ethereums throughput may increase to 100,000 in the future when they have implemented their sharding upgrade. But as I said earlier, Hedera can theoretically already do that natively.
Hedera was using the EVM for supporting smart contracts aswell, it is open sourced. Transactions of smart contracts are a bit more complicated so let’s give him a benefit of a doubt and take his claim that Hedera could only do a maximum of 10 TPS when handling smart contracts, in 2019. This was at a time when Ethereum was still running a PoW algorithm.
So you take something that was designed for ~10 TPS, put it on a Ledger that can do 10,000 TPS and you expect it to go faster? If the EVM can’t handle more than ~10 TPS, ofcourse it won't be able to read information faster just because you put it on a faster network, the buffer will overflow!
For a more human readable analogy: If your car is from the 40's and the engine can only go up to 30km/h, that is your maximum speed, regardless if you are driving on a dirt road or the autobahn.
So the core argument is that 10,000 TPS is only for, his words, “dumb” transactions, for example crypto currency transactions. That is true. It even says so in the footnote. For other services it could vary. But isn’t that like comparing apples with oranges? The fact still remains, when you compare apples with apples, i.e. with only crypto currency transactions:
Bitcoin: ~7 TPS
Ethereum: ~14-27 TPS
Hedera: 10,000 TPS
You can just try it out in practice and you will see for yourself how much faster and cheaper it is to make a transaction on Hedera, compared with Bitcoin and Ethereum.
Ofcourse, Smart Contracts are important and Hedera should aim to improve the performance for that. Fast forward to 2022 and Hedera has done precisely that, with Smart Contracts 2.0. Now it can handle 350 smart contract transactions per second, thottled.
Note: Smart Contracts 2.0 is based on the open source project Hyperledger BESU so this should not be unique to Hedera.
But what about the state proofs? State proofs are records on a state of the network at any given time. For example, if you want to check the current status of a transaction. However, there are many other more advanced use-cases.
The reason I took it up is because the author questions why state proofs (transaction records) are excluded from the 10,000 TPS claim, zooming in on the footnote:
As state proofs can be used for different services, the transaction throughput for state proofs will also vary, depending on what service you are querying. However, there are no limits for receipts and should therefore be just as fast as crypto currency transactions.
Lastly, this quote made me chuckle:
“…Btw, did I mention these state proofs costs money to receive?”
You mean the $0.0001 transaction fee? Yeah I guess you may have to pay for that. Receipts are free though. But he is really being the pot calling the kettle black when he complains about transactions fees while defending Bitcoin and Ethereum.
The rest of the article goes on about addressing some theoretical vulnerabilities with the Hedera network and has nothing to do with transaction throughput. Some may have had a point back then but since this article is almost 5 years old, those vulnerabilities may not be relevant anymore. Besides, it’s not like Bitcoin or Ethereum don’t have their own share of theoretical vulnerabilities.
Conclusion
From an investing standpoint, Eric made a good decision to stay with his Bitcoin and Ethereum, with more than 500% increase in value for Bitcoin and 1400% increase for Ethereum.
It’s called currency for a reason. It is supposed to flow through an economic system. But people rather hoard BTC and ETH as speculative commodities instead.
But how long will it last? When considering the technical benefits of Hedera, it makes you wonder why HBAR hasn’t caught up in the crypto currency market yet. If I was a web3 developer and going to choose a network for my apps, Hedera seems ideal with it’s speed, security, stability and predictable costs. I would at least choose anything else but Bitcoin.
My expectations is that Bitcoin is going to be netscaped in a few years due to their unsustainable infrastructure. Unless Bitcoins perceived value increases significantly every year, miners will go broke to keep the network alive. I don’t think Lightning network will be enough to save it.
Ethereum on the other hand, may live on for a while longer. It doesn’t have the same maintenance costs as Bitcoin. They are working on making the infrastructure faster, but I believe they have to sacrifice some security for that to happen. I wouldn’t be surprised if in the future we can read about some disastrous attack and billions of dollars worth of assets are stolen.