News | October 21, 2015


Store and Process Massive-Scale Graphs in Real Time

Neo4j on POWER8 offers 56 TB of extended memory, drastically increasing the size at which real-time graph queries are possible.

Neo4j on IBM POWER8 is the result of a joint effort between Neo4j and IBM engineering to provide the world’s most scalable graph database platform capable of storing and processing graphs of extremely large size all in-memory – shattering all previous real-time scalability limits.

As the world’s leading graph database, Neo4j helps business leaders not only manage larger volumes of data but generate insights from their data relationships, and they apply those insights in real time at the point of touch – a task that requires the ability to process massive volumes of data at scale, which IBM Power Systems are designed to handle.

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.

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

IBM Power Systems Built with POWER8

IBM Power Systems are specifically designed to capture and manage data from a variety of sources and put that data to work in your enterprise – including real-time graph processing.

Only Power Systems provide up to 56 terabytes of extended memory space with CAPI (Coherent Accelerator Processor Interface) flash architecture, accommodating graphs of huge size and scale.

IBM Power Systems built on POWER8 processors are optimized for a broad range of data workloads, including graph data. IBM Power Systems deliver cost-effective, high-performing and agile IT infrastructure so enterprises can leverage the most advantage from their diverse data sources.

Highlights of the POWER8 advantage:

  • Processors:
    • 4x more hardware threads per core (versus an x86 core)
    • 96 threads on a 12-core chip
    • Up to 1536 threads per system
    • Better Simultaneous Multi-threading (SMT) performance than Hyper-threading (HT)
  • Memory:
    • 4x more memory bandwidth (versus an x86 core)
    • Up to 16 TB of DRAM
    • Consistent latency as DIMMs are added
    • 192 GBs of sustained memory bandwidth per scale-out socket
    • Up to 230 GBs per second of memory bandwidth for enterprise-class server
  • Cache:
    • 4x more cache (versus an x86 core)
    • Up to 231MB cache per socket (for a 12-core chip)
  • I/O Bandwidth:
    • 96 GBs per second of peak bandwidth
    • 4x more bandwidth than previous generation (POWER7)

Source: Neo Technology