MiniBase is an embedded KV storage engine, it's quit simple, not for production env, just for better understand HBase or other LSM-related index algorithm.
In MiniBase, we use those basic algorithm and data structure:
- BloomFilter: it can helps a lot when filtering much useless IO.
- ConcurrentSkipListMap: Yeah, it's quite suitable when designing memstore. It can maintian an sorted key value set in high concurrency scenarios.
- LSM Index Algorithm: the memstore part and disk store part.
Configconf = newConfig().setDataDir(dataDir).setMaxMemstoreSize(1).setFlushMaxRetries(1)
.setMaxDiskFiles(10);
MiniBasedb = MiniBaseImpl.create(conf).open();
// Putdb.put(Bytes.toBytes(1), Bytes.toBytes(1));
// ScanIter<KeyValue> kv = db.scan();
while (kv.hasNext()) {
KeyValuekv = kv.next();
//...
}git clone git@github.com:openinx/minibase.git
mvn clean package