class: title-slide # Revisiting Transactional Statistics of
High-scalability Blockchains
Daniel Perez
Imperial College London
Jiahua Xu
University College London
Benjamin Livshits
Imperial College London
October 29th, ACM Internet Measurement Conference 2020
Slides available at: http://bit.ly/transactional-throughput
--- # Overview 1. Background 2. Data overview 3. Main results 1. EOS 2. Tezos 3. XRP 4. Takeaways --- class: middle, title-slide # Background --- # Blockchain overview .pull-left-40[ * Append-only data structures * Unit-of-work is a block * Each block has many transactions * Each transaction can contain several actions * Block producers (miners) need to reach consensus ] .pull-right-60[
] --- class: full-table # Blockchain scalability * Scalability often discussed in terms of transactions per second * Average transactions per second can be computed by
blocks/second x transactions/block Platform | Block time | Transactions/block | Transaction/second :---------:|----------------:|-------------------:|------------------: Visa | - | - | 65,000 Bitcoin | 10 minutes | 3,500 | 6 -- Tezos | 1 minute | 2,400 | 40 EOS | 0.5 second | 2,000 | 4,000 XRP | ? | ? | 65,000 .caption[ Advertised limits of different platforms ] --- class: middle, title-slide # Data overview --- class: full-table # Data collected .pull-left-40[ * Data from October 1, 2019 to April 30, 2020 * Grouped in 6-hour interval for TPS computation ] .pull-right-60[ Platform | Number of blocks | Number of transactions | Average TPS | Max TPS :---------:|-------------------:|-----------------------:|-------------:|--------: EOS | 36,133,709 | 631,445,236 | 34 | *136* Tezos | 301,822 | 7,890,133 | 0.43 | *0.57* XRP | 4,753,965 | 271,546,797 | 15 | *56* ] --
Data and measurement framework open-sourced: https://github.com/danhper/blockchain-analyzer --- class: full-table # Types of transactions
EOS
Tezos
XRP
Category
Action name
%
Operation kind
%
Transaction type
%
Peer-to-peer transactions
Transfer
96.2
Transaction
21.4
Payment
36.9
EscrowFinish
0.0
Account actions
newaccount
0.0
Reveal
0.0
TrustSet
1.2
bidname
0.0
Origination
1.3
AccountSet
0.1
Other actions
delegatebw
0.0
Endorsement
76.6
OfferCreate
59.1
undelegatebw
0.0
Delegation
0.6
OfferCancel
2.7
buyrambytes
0.0
Reveal nonce
0.1
EscrowCreate
0.0
Others
3.8
EnableAmendment
0.0
.caption[ Distribution of actions executed on each blockchain ] --- class: middle, title-slide # EOS results --- # EOS Throughput over time .pull-left-40[ * Change of trend in November 2019 * Due to airdrop of EIDOS token * Caused EOS to enter congestion mode ] .pull-right-60[ ![EOS throughput over time](./eos-chart-area.png) ] --- # EIDOS token aidrop .pull-left-50[ * Very simple airdrop 1. Send EOS to a contract 2. Contract sends back the EOS and some amount of its token * Originator of the airdrop is anonymous but tone is aggressive * Likely a voluntary spam attack * Caused many developers to quit EOS ] .pull-right-50[ ![EIDOS tweet](./eidos-tweet.png) ] ??? According to recent report, developers working on platform fell by 85% https://decrypt.co/32217/billion-dollar-blockchain-eos-in-trouble-research --- # EOS Throughput before spam attack .pull-left-40[ * Used a lot for betting games * Used for some porn website payments * Fairly stable and varied usage ] .pull-right-60[ ![EOS throughput over time](./eos-old-chart-area.png) ] --- # Wash-trading on EOS * WhaleEx was most active exchange at time of analysis * Claims to be the largest decentralized exchange in the world * In *75%* of the trades, the buyer and seller were the same user * The exact same amount was sent back-and-forth * Aggregated trades for all top users resulted in 0 value transfer --- class: middle, title-slide # Tezos results --- # Tezos Throughput over time .pull-left-40[ * Very stable trend * Vastly dominated by endorsement - consensus related * Regular spikes in number of transactions * Transaction spikes due to block producer payouts - consensus related ] .pull-right-60[ ![Tezos throughput over time](./tezos-chart-area.png) ] --- class: middle, title-slide # XRP results --- # XRP Throughput over time .pull-left-40[ * Dominated by `OfferCreate` * Spike around November and December due to spam attack * Stable throughput from January and on ] .pull-right-60[ ![XRP throughput over time](./xrp-chart-area.png) ] --- # I Owe You (IOU) in XRP .pull-left-50[ * Anybody can emit an IOU * User can establish trust line to accept IOUs from other users * If the IOU issuer cannot pay, the IOU has no value ] .pull-right-50[ ![Tweet about XRP IOU](./tweet-xrp.png) ] --- # Value in XRP transactions .pull-left-50[ * Less than *6%* of the transactions had an economic value * Only 0.2% of the transactions were successful trades * Only 3.3% of the transactions were payments with an economic value ] .pull-right-50[ ![XRP transactions value](./xrp-sankey.png) ] --- # Valuable payments on XRP .pull-left-40[ * Vast majority of valuable payments are in XRP * Average of 586 million XRP traded per day * Exchanges account for a large portion of these payments ] .pull-right-60[ ![XRP valuable transactions](./xrp-cashflow.png) ] ??? BTC has > 30 billion trading volume per day --- # Takeaways * All three blockchains are still very far from their potential * Current TPS very far away from claimed maximum * Most transactions are have no economic value * *96%* of EOS transactions from valueless airdrop token * *76%* of Tezos transactions to maintain consensus * *94%* of XRP transactions have no economic value * Low or absent fees tend to attract a lot of spam * XRP has dealt with spam attacks much better than EOS, likely because of the difference in fee model