Neo4j Graph Database 3.3 Release: Everything You Need to Know

As part of the Neo4j Graph Platform announced at GraphConnect New York last week, we are excited to announce the general availability release of the Neo4j Graph Database version 3.3.

At the heart of the Graph Platform is the native graph database itself, which has a number of release-defining features to satisfy stakeholders across the IT organization.

Let’s take a closer look at the improvements and upgrades new to the Neo4j Database 3.3:

What’s New in Neo4j 3.3 – All Editions:

Cypher and Write Performance

Write performance is 55% faster than Neo4j 3.2 and nearly 350% faster than version 2.3, making it possible to ingest more data in a shorter time. Transactional writes benefit from new native indexes, which replace Neo4j’s Lucene-based indexes for numeric properties. You’ll recall that in 3.2, we went native for label indexes which accounted for most of the performance boost in that release.

Neo4j’s ACID characteristics, as with the previous indexes, encompass the indexes in addition to the data. Bulk writes have also received a major boost: reducing the memory footprint by up to 40% and leveraging virtual memory in RAM-constrained environments.

Cypher performance has also improved. The new Cypher interpreter is now 40-70% faster according to internal tests.

Data Import, Integration & ETL

Neo4j’s data import functionality uses 40% less memory and adds page caching for large imports.

The new (pre-release) Neo4j Data Lake Integrator toolkit supports the import and export of HDFS files into graphs and back again.

Also in pre-release is the new Neo4j ETL that simplifies the process of morphing data from relational database management systems (RDBMS) into graph data. This process reveals previously invisible long-path relationships hidden within structured data.

RDBMS-to-graph and bulk importing improvements let you hit the ground running with Neo4j.

Upgrades for Graph Analytics

As part of the Graph Analytics component of the native Graph Platform, data scientists now enjoy the new impressively fast graph algorithms library – included as part of the APOC library – that now ships with Neo4j Database 3.3.

Use graph algorithms to reveal unseen patterns in your connected data.

Also of significant note is Cypher for Apache Spark (CAPS) which not only binds the ease of writing Cypher with a massively scalable, in-memory graph analytics processor, but its queries also return subgraphs. This makes Cypher composable and allows users to chain graph queries in sequence to carry out complex algorithmic logic. Graphs can now be saved in both Neo4j and as snapshot files in HDFS.

What’s New in Neo4j 3.3 – Enterprise Edition Only Features:

Neo4j Desktop

The leading Enterprise Edition feature of the 3.3 release is the introduction of Neo4j Desktop. This package is the new developers’ mission control console.

It provides a free local instance of Neo4j Enterprise Edition for development and includes access to Neo4j graph visualization and development tools and an installer for the APOC and graph algorithm libraries.

The Neo4j Desktop package is your mission control for organizing and managing all of your Neo4j projects.

Neo4j Database Kernel Improvements

Numerous improvements to the database kernel expand database availability during upgrades.

Neo4j Enterprise Edition now allows key configuration parameters to be changed on the fly, without needing to recycle a database instance. Local Schema Locks automatically narrow the scope of locks, avoiding the need for the database to take a global schema lock when creating or changing a schema object or constraint.

The database now also offers a cache-hit ratio, available at both a query and database level, as a new metric for helping to size one’s database cache.

Scalability Improvements: Load Balancing, Security & More

Other Enterprise Edition feature upgrades and improvements worth noting include:

• Intra-server encryption keeps all server-to-server transmissions safe, from data center to data center and across cloud zones – all managed in Bolt.
• The Bolt driver’s new least-connected load balancing replaces round-robin selection in order to maintain high cluster throughput under high demand conditions.
• Also, node ids are fetched in batch, removing an otherwise pesky bottleneck revealed when clusters are under heavy activity.
• Clusters now support IPv6 which sets the stage for trillions of device and IP connections.

Conclusion

As you can see, with the advent of the Neo4j Graph Platform the core graph database has not only improved within itself but also in relation to all other components of the platform.

We’re confident that the 3.3 release of the Neo4j Graph Database will by far be everyone’s favorite release to date. If you haven’t already, make the upgrade to 3.3 and see why Neo4j is the #1 platform for connected data.