Moving off of Neo4J - Githubissues

telstra / open-kilda

OpenKilda is an open-source OpenFlow controller initially designed for use in a global network with high control-plane latency and a heavy emphasis on latency-centric data path optimisation.

Apache License 2.0

78 stars 53 forks source link

Moving off of Neo4J #940

Closed carmine closed 3 years ago

carmine commented 6 years ago

ONLY Neo4J EE supports scaling and HA. Proof: https://neo4j.com/docs/operations-manual/current/ha-cluster/basic-installation/

Follow these steps in order to configure a Neo4j HA cluster:
Download and install the **Neo4j Enterprise Edition** on each of the servers to be included in the cluster.

surabujin commented 6 years ago

@carmine what benefits do we have from neo4j?

sergii-iakovenko commented 4 years ago

Created two PoC with benchmarks: #2747 (Persistence / storage replacement PoC for Ferma + OrientDB) and #2770 (PCE replacement PoC for Ferma + OrientDB). Git branches: neo4j-replacement-poc and neo4j-replacement-poc-2.

sergii-iakovenko commented 4 years ago

The next steps:

Develop a solution with clustered OrientDB:
- Run clustered OrientDB in dev & all-in-one enviroments.
- Options for GUI / client to access data in the storage.
- Data-model: is Ferma + data gateway pattern the only option? How to perform partial data reads / entity views?
- Multi-threaded benchmark tests: deadlocks, locks, read-only access.
- Options for PCE: native, Gremlin Server, Kilda in-memory. The requirements the same as https://github.com/telstra/open-kilda/pull/2770
Other storages?

sergii-iakovenko commented 3 years ago

The migration is completed. PCE improvements and minor issues related to OrientDB will be addressed separately.