sqlChain - The Blockchain as a SQL Layer
sqlChain is a compact SQL layer that runs on top of bitcoind (and some altcoins). It extends the query options on the blockchain with a priority placed on low storage overhead. It provides multiple API (compatible) interfaces:
- Insight API (plus some extensions, like /api/closure)
- Blockchain.info API (including websocket)
- RPC via POST, GET urls
- Web Interface (demo of integrating with API backend; only hints at what you can do)
- Electrum Server (old, still untested, needs TLC)
sqlChain currently consists of three daemon programs.
- sqlchaind - monitors bitcoind and updates a mysql database
- sqlchain-api - provides multiple API interfaces over the mysql database
- sqlchain-electrum - runs a private electrum server over the sqlchain-api layer
Recent Updates
(v0.3.0) (In progress) Migration to Python3.
Also now added a Roadmap Project to Github. So if you are interested in the future of this project and a renewed committment to taking it into the next decade then have a look.
(v0.2.9)
Made API modules config selectable or Insight compatible if none set. Now a supplied API can be customized and integrated by editing the config file. This makes it easier to build web apps that need specific API calls. sqlchain-config adds all 5 supplied APIs by default. If upgrading you may need to edit the config file to add ones you use.
New config option called max_blks that limits how many blocks are kept in SQL db. This is useful when you primarily want to process realtime or not very old data. Used with a pruning node it indexes address/tx data back as far as max_blks, which is something you cannot do on a pruned node normally as txindex cannot be used. It also means you can start from a given block (set block in config) and not have to sync SQL db from Genesis; this allows a quick start up in some cases. Also fixed bugs in sync api.
(v0.2.5)
Database sync code updated for SQL transactional engines. Tested with MariaDB using the RocksDB engine. This engine has some nice features but the main ones of interest here are storage size reduction and indexing (instead of Btree) more suited to high entropy keys (tx,address ids). In my tests RocksDB was not much faster initially but didn't drop in speed so much as DB size grows. It's a bit early to fully recommend but initially it looks like a nice option. I'll update the install guide with RocksDB steps (soonish).
Added bech32 address support (p2wpkh and p2sh). This requires a database upgrade and along with other changes the best option is to re-sync the blockchain. sqlchain will stop if it detects an old db and if re-sync is not possible then reverting to pre v0.2.5 is best.
Now supports multiple blockchains and testnet variants. Currently Litecoin, Dogecoin and Reddcoin have been added as test cases (with demo pages) and I hope to add a few more before long. Each coin requires it's own daemon process but sqlchain-config (sqlchain-init replacement) now takes advantage of systemd "instances" so that several can coexist. This means the systemctl commands are now like systemctl start sqlchain@bitcoin, and similarly for other coins. There is only one [email protected] and it creates variant instances for each coin described by it's cfg.
Upstart (Ubuntu 14.04) support has been removed - it probably works fine but the setup process now only automates Ubuntu 16.04 (systemd) and newer platforms.
As part of new altcoin support there is now an "overlay" feature where custom modules can be loaded based on cointype or for extending an api. SQL schema can be overridden likewise based on cointype or db name. If you have a custom schema you can have it initialized by sqlchain-config simply by using a matching custom db name. Both these options allow customizing and extending code while easing the burden of merging updates.
New unit tests have been added, see the README in the tests directory. Many bugs fixed and api behaviour improved as a result.
See re-organized docs directory for more detailed info on adding new altcoins and running with alternative database engines.
Supported Features (with more tested history)
- testnet and segwit - decodes and stores witness data but not much of an segwit api yet
- pruning - since block data is parsing into mysql you can remove old blocks and save ~50% disk space with no loss in api
- no-sig option - can drop signature and witness data to further reduce disk space for uses not requiring proofs and raw tx data
- external blobs - most signature, witness and script data is offloaded to blobs exteral to mysql, giving finer control (losing indexibility)
- split blobs, s3 ready - blobs are split in 5GB chunks, allows mapping older tx data to cheaper offline storage like Amazon s3
- blkdat mode - direct reading of bitcoind block files allows much faster initial sync as sqlchain can read while bitcoind is syncing
- blkbtc - utility to block on/off node network traffic to allow more cpu for sqlchain to catch up, or limit disk used by syncing
- sqlchain-config - dialog based utility to ease setup and generate optimal config files
sqlchain is still Beta software, under sporadic active development.
sqlchain-electrum has not received much love over the last 2 years but I do plan to get it caught up and functioning again.
Try It Out
You can try it on Testnet and it doesn't take much time or resources. Even a 1vCPU (1.5 cents/hour) Vultr instance can run it quite well. You can snapshot the instance and only run as needed. On this VPS Testnet sync'd in 45 minutes and used ~ 12 GB. It takes ~1.5 days to sync mysql data to block 1156000. The first block with segwit txs seems to be 834624.
TODO
See the new Roadmap Project.
Donations supporting further development are welcome.