Neo4j on IBM POWER8: Scale Massive Graphs Like Never Before


Learn More about Scaling Massive Graph Data with Neo4j on IBM POWER8 TechnologyopenCypher and the Neo4j 2.3 release weren’t the only big announcements Emil Eifrem made during his keynote address at GraphConnect San Francisco.

The Neo4j team is also proud to unveil Neo4j on IBM POWER8.

Just as we are working hard to make Neo4j the best, fastest and most scalable platform for running graphs on standard on-premise and cloud infrastructure, we saw an opportunity – together with IBM – to advance the art of the possible by marrying the power of Neo4j to IBM’s quite impressive next generation of POWER hardware: POWER8.

This joint effort between Neo4j and IBM engineering provides an offering at the ultra-high end of the scale, where very big data meets real-time graph processing.

For companies that prefer to scale out, or simply don’t have a 56 TB (!!) in memory graph, Neo4j now (as of Neo4j 2.3) officially spans the entire range of IBM POWER8 Linux servers, including “commodity” servers that are both powerful and cost-efficient.

The Opportunity: Massive In-Memory Graph Processing for Real-Time Scale


A key challenge for graph processing at scale is how to handle the sheer size of today’s emerging datasets without compromising real-time (OLTP) capabilities.

Scaling horizontally is one obvious and common way to scale, but if your application is real time, then hopping all over your cluster within a network (or excessively relying on disk) isn’t the best recipe for fast, predictable reads.

The best way to get fast and predictable performance is to take advantage of the ample memory available on today’s systems. Neo4j lets you do just that.

However, Neo4j + POWER8 takes it to another level: with systems offering up to 12 TB of RAM and 40 TB of CAPI-attached flash accessible at near-memory throughputs and speeds.

IBM Power Systems Built with POWER8


IBM Power Systems built with POWER8 processors provide up to 56 terabytes of extended memory space with CAPI flash architecture, greatly increasing the scale of graphs that can be stored and processed in-memory.

The POWER8 is a massively multi-threaded processor. Each core is capable of handling eight hardware threads simultaneously for a total of 96 threads executed simultaneously on a 12-core chip.

The processor makes use of large amounts of on- and off-chip eDRAM caches, and on-chip memory controllers enable high bandwidth to memory and system I/O.

In addition, the coherent accelerator processor interface – or CAPI – is a new capability that is part of the POWER8 chip. CAPI flash architecture allows off-chip accelerators to have coherent access to POWER8 memory just like traditional power cores.

CAPI-attached flash technology introduces a new tier of memory that combines lightning quickness with massive capacity, greatly increasing the size at which real-time graph queries are possible.

Neo4j + POWER8 = The Perfect Solution


The combination of Neo4j’s native graph processing and storage and POWER8’s in-memory vertical scalability is a natural convergence. Neo4j on POWER8 makes it possible to store and process massive-scale graphs in real time – a problem that was simply unsolvable only yesterday.

Neo4j officially supports Linux on IBM Power Systems beginning with Neo4j 2.3. Additionally, Neo Technology and IBM are now accepting beta test participation for POWER8 with CAPI flash, which will begin Q4 2015. Please contact us to participate.

Together, the strength and performance of Neo4j plus the scalability and speed of POWER8 will provide unrivaled graph application performance.

Conclusion


The union of Neo4j on POWER8 presents a golden opportunity for today’s enterprise leaders and application developers.

With the power of these two technologies, data professionals overcome any and all scalability challenges with a graph database that extracts meaningful, real-time insights from data relationships and the hardware that allows it to scale to massive proportions.

Not only does Neo4j on POWER8 deliver real-time graph processing, but the robustness of IBM hardware allows enterprises to scale their graph data to levels never previously imagined.


Click below to read the Neo4j on IBM POWER8 data sheet and discover how your enterprise can tap into real-time graph processing that scales higher than ever before.