Presto Ethereum Connector

Unleash the Power of Presto Interactive SQL Querying on Ethereum Blockchain

Introduction

Presto is a powerful interactive querying engine that enables running SQL queries on anything -- be it MySQL, HDFS, local file, Kafka -- as long as there exist a connector to the source.

This is a Presto connector to the Ethereum blockchain data. With this connector, one can get hands on with Ethereum blockchain analytics work without having to know how to play with the nitty gritty Javascript API.

Prerequisites

Have an Ethereum client that you can connect to. There are 2 options:

Run Geth or Parity locally.
Use Infura, a hosted Ethereum client in the cloud.

Note

Specify a block range where you can (e.g. WHERE block.block_number > x AND block.block_number < y, or WHERE transaction.tx_blocknumber > x AND transaction.tx_blocknumber < y, or WHERE erc20.erc20_blocknumber > x AND erc20.erc20_blocknumber < y). Block number is the default and only predicate that can push down to narrow down data scan range. Queries without block ranges will cause presto to retrieve blocks all the way from the first block, which takes forever.

Usage

Install Presto. Follow the instructions on that page to create relevant config files.
By the end of this step, your presto installation folder structure should look like:
```
├── bin
├── lib
├── etc
│   ├── config.properties
│   ├── jvm.config
│   └── node.properties
├── plugin
```
Install Presto CLI
Clone this repo and run mvn clean package to build the plugin. You will find the built plugin in the target folder.

Load the plugin to Presto
a. Create the ethereum connector config inside of etc.
$ mkdir -p etc/catalog && touch etc/catalog/ethereum.properties
Paste the following to the ethereum.properties:

connector.name=ethereum

# You can connect through Ethereum HTTP JSON RPC endpoint
# IMPORTANT - for local testing start geth with rpcport
# geth --rpc --rpcaddr "127.0.0.1" --rpcport "8545"
ethereum.jsonrpc=http://localhost:8545/

# Or you can connect through IPC socket
# ethereum.ipc=/path/to/ipc_socketfile

# Or you can connect to Infura
# ethereum.infura=https://mainnet.infura.io/<your_token>

b. Copy and extract the built plugin to your presto plugin folder

$ mkdir -p plugin/ethereum \
  && cp <path_to_this_repo>/target/presto-ethereum-*-plugin.tar.gz . \
  && tar xfz presto-ethereum-*-plugin.tar.gz -C plugin/ethereum --strip-components=1

By the end of this step, your presto installation folder structure should look like:

  ├── bin
  ├── lib
  ├── etc
  │   ├── catalog
  │   │   └── ethereum.properties
  │   ├── config.properties
  │   ├── jvm.config
  │   └── node.properties
  ├── plugin
  │   ├── ethereum
  │   │   └── <some jars>

There you go. You can now start the presto server, and query through presto-cli:

$ bin/launcher start
$ presto-cli --server localhost:8080 --catalog ethereum --schema default

Use Cases

Inspired by An Analysis of the First 100000 Blocks, the following SQL queries capture partially what was depicted in that post.

The first 50 block times (in seconds)

SELECT b.bn, (b.block_timestamp - a.block_timestamp) AS delta
FROM
(SELECT block_number AS bn, block_timestamp
FROM block
WHERE block_number>=1 AND block_number<=50) AS a
JOIN
(SELECT (block_number-1) AS bn, block_timestamp
FROM block
WHERE block_number>=2 AND block_number<=51) AS b
ON a.bn=b.bn
ORDER BY b.bn;

Average block time (every 200th block from genesis to block 10000)

WITH
X AS (SELECT b.bn, (b.block_timestamp - a.block_timestamp) AS delta
    FROM
        (SELECT block_number AS bn, block_timestamp
        FROM block
        WHERE block_number>=1 AND block_number<=10000) AS a
    JOIN
        (SELECT (block_number-1) AS bn, block_timestamp
        FROM block
        WHERE block_number>=2 AND block_number<=10001) AS b
    ON a.bn=b.bn
    ORDER BY b.bn)
SELECT min(bn) AS chunkStart, avg(delta)
FROM
(SELECT ntile(10000/200) OVER (ORDER BY bn) AS chunk, * FROM X) AS T
GROUP BY chunk
ORDER BY chunkStart;

Biggest miners in first 100k blocks (address, blocks, %)

SELECT block_miner, count(*) AS num, count(*)/100000.0 AS PERCENT
FROM block
WHERE block_number<=100000
GROUP BY block_miner
ORDER BY num DESC
LIMIT 15;

ERC20 Token Movement in the last 100 blocks

SELECT erc20_token, SUM(erc20_value) FROM erc20
WHERE erc20_blocknumber >= 4147340 AND erc20_blocknumber<=4147350
GROUP BY erc20_token;

Describe the database structure


SHOW TABLES;
Table
-------------
block
erc20
transaction

DESCRIBE block; Column | Type | Extra | Comment

DESCRIBE transaction;

Column | Type | Extra | Comment


### Web3 Functions
In addition to the various built-in [Presto functions](https://prestodb.io/docs/current/functions.html), some web3 functions are ported so that they can be called inline with SQL statements directly. Currently, the supported web3 functions are
1. [fromWei](https://github.com/ethereum/wiki/wiki/JavaScript-API#web3fromwei)
1. [toWei](https://github.com/ethereum/wiki/wiki/JavaScript-API#web3towei)
1. [eth_gasPrice](https://github.com/ethereum/wiki/wiki/JavaScript-API#web3ethgasprice)
1. [eth_blockNumber](https://github.com/ethereum/wiki/wiki/JavaScript-API#web3ethblocknumber)
1. [eth_getBalance](https://github.com/ethereum/wiki/wiki/JavaScript-API#web3ethgetbalance)
1. [eth_getTransactionCount](https://github.com/ethereum/wiki/wiki/JavaScript-API#web3ethgettransactioncount)

### Troubleshooting

* You must use python2. You will get invalid syntax errors if you use Python3.

-> bin/launcher start File "/your_path/presto-server-0.196/bin/launcher.py", line 38 except OSError, e: ^ SyntaxError: invalid syntax


* Use Java 8 only. You might get the following errors if you use the wrong Java version.

Unrecognized VM option 'ExitOnOutOfMemoryError' Did you mean 'OnOutOfMemoryError='? Error: Could not create the Java Virtual Machine. Error: A fatal exception has occurred. Program will exit.