3 ms·

Show HN: TerminusDB 1.1 – Big Babushka' OSS graph DB that stores data like Git

3 points by LukeEF 7y ago

LukeEF 7y ago
TerminusDB 1.1 - the Big Babushka! So happy to put this out. It could almost be a 2.0, but given that we are a young DB, we want to save those release numbers for future exciting developments (I did see another DB on release 0.50.0 - I won’t name names, but come on folks, give us a 1.0!). The major development is our new storage backend, terminus-store. With the introduction of the new store, TerminusDB takes a radical departure from traditional database architectures. Our approach has three parts: 1. We have a graph database with strong schema to retain simplicity and generality 2. The graph is implemented using succinct immutable data structures which enable more sparing use of main memory resources. 3. We adopt a delta encoding approach to updates (‘like git, but for data’) which provides the whole suite of revision control features: branch, merge, squash, rollback, blame, and time-travel facilitating CI/CD approaches on data. This final point is crucial - with TerminusDB we can now use advanced CI/CD workflows in data operations. The large impact of git on the software development world can now be envisioned in the world of data. This is made possible by synergies between an immutable layered approach and the succinct data structure approach that we’ve used for encoding. The DevOps revolution becomes the DataOps revolution. In addition to the new backend, we have a re-imagined console, our query language WOQL has been extended and, importantly, we have a Python Client. We are the database for data people and there are no more data people than Pythonistas.
bionhoward 7y ago
Interesting concept and would use it if it were Serverless and had a Gremlin or Cypher client. Right now it’s hard to see how this is a “Graph DB” as the API has no pattern matching and traversals (but will keep an eye on it) ... without traversals it’s more like a version controlled MongoDB than a graph database. Add something like cypher MATCH and something like Gremlin traversals or ArangoDB FOR, and add full text search, host it with multitenancy, and rule the world and take my money

3 points by LukeEF 7y ago

TerminusDB 1.1 - the Big Babushka! So happy to put this out. It could almost be a 2.0, but given that we are a young DB, we want to save those release numbers for future exciting developments (I did see another DB on release 0.50.0 - I won’t name names, but come on folks, give us a 1.0!). The major development is our new storage backend, terminus-store. With the introduction of the new store, TerminusDB takes a radical departure from traditional database architectures. Our approach has three parts: 1. We have a graph database with strong schema to retain simplicity and generality 2. The graph is implemented using succinct immutable data structures which enable more sparing use of main memory resources. 3. We adopt a delta encoding approach to updates (‘like git, but for data’) which provides the whole suite of revision control features: branch, merge, squash, rollback, blame, and time-travel facilitating CI/CD approaches on data. This final point is crucial - with TerminusDB we can now use advanced CI/CD workflows in data operations. The large impact of git on the software development world can now be envisioned in the world of data. This is made possible by synergies between an immutable layered approach and the succinct data structure approach that we’ve used for encoding. The DevOps revolution becomes the DataOps revolution. In addition to the new backend, we have a re-imagined console, our query language WOQL has been extended and, importantly, we have a Python Client. We are the database for data people and there are no more data people than Pythonistas.

bionhoward 7y ago

Interesting concept and would use it if it were Serverless and had a Gremlin or Cypher client. Right now it’s hard to see how this is a “Graph DB” as the API has no pattern matching and traversals (but will keep an eye on it) ... without traversals it’s more like a version controlled MongoDB than a graph database. Add something like cypher MATCH and something like Gremlin traversals or ArangoDB FOR, and add full text search, host it with multitenancy, and rule the world and take my money