IBM Data Engine for NoSQL is an integrated platform for large and fast growing NoSQL data stores. It builds on the CAPI capability of POWER8 systems and provides super-fast access to large flash storage capacity. It delivers high speed access to both RAM and flash storage which can result in significantly lower cost, and higher workload density for NoSQL deployments than a standard RAM-based system. The solution offers superior performance and price-performance to scale out x86 server deployments that are either limited in available memory per server or have flash memory with limited data access latency.

Benefits

IBM Data Engine for NoSQL is an innovative solution that can:

  • Provide lower total cost of ownership for NoSQL infrastructure with high levels of consolidation as compared to traditional approaches
  • Deliver from 2TB to 57TB of a new tier of super-fast storage with one POWER8 processor-based server with Coherent Accelerator Processor Interface (CAPI)-attached flash storage and no noticeable performance degradation
  • Enable deployment of larger NoSQL databases on a single or smaller number of servers as well as the ability to consolidate multiple NoSQL databases on a smaller footprint
  • Supports explosive growth of new mobile and social apps enabled by NoSQL with lightning-fast access to data with simplified database management and operation.

Data management challenge

The explosive growth of social applications, big data, mobile access and cloud computing is generating large volume and variety of data and is changing the way that applications are developed. The need to manage data at an unprecedented scale not seen before is driving the need for new data models to store, manage and access data. This has led to a new class of data stores which fall under the broad category of Not Only SQL (NoSQL) databases.

NoSQL databases are increasingly considered a viable alternative to relational databases that offers a flexible data model and exceptionally fast access to data. Typical NoSQL implementations run entirely in-memory, or rely heavily on memory as a cache. Typical scale-out deployments can therefore become expensive and hard to manage. Flash memory based enhancements to address this scaling challenge have lower costs but the increase in data access latency leads to challenges in meeting application performance requirements.


Innovative solution for NoSQL database deployment

A key innovation in IBM POWER8’s open architecture is the Coherent Accelerator Processor Interface (CAPI). CAPI provides a high bandwidth, low latency path between external devices, the POWER8 core, and the system’s open memory architecture. The IBM Data Engine for NoSQL leverages the CAPI capability to provide a high throughput, low latency connection to flash storage to create a unique super-fast storage tier that addresses the scaling issues of NoSQL deployments. The design allows for the processor main memory to provide fast response times that applications require by utilizing main memory and flash storage to cache or hold “hot” data. The Flash storage is attached using a CAPI adapter card to provide a high bandwidth, low latency path between the processor and the flash. Providing the POWER8 processors direct access to both DRAM and Flash allows application software to adjust memory and flash usage ratios to optimize performance and cost based on the specific service level agreements.

Deployment options

Two configuration options are available for IBM Data Engine for NoSQL

Internal integrated CAPI-Flash adapter based configuration – Delivers from 2TB to 8TB of super-fast flash storage with a single POWER8 processor-based server with Flash modules integrated with the CAPI card.

 

External IBM FlashSystem based configuration - Deliver up to 57TB of super-fast flash storage with one POWER8 processor-based server with CAPI card attached to an external IBM FlashSystem

Solutions

The following solutions are optimized to leverage IBM Data Engine for NoSQL platform

RedisLabs

Redis Enterprise Cluster by RedisLabs, Inc. leverages the IBM Data Engine for NoSQL platform to offer a unique solution that cuts down the deployment cost of Redis, while maintaining extremely high read/write operations per second at sub-millisecond latency to ensure application response time. Redis Enterprise Cluster makes it easy to adjust the volume of RAM versus Flash that is used, as a means to optimize for cost savings and monitor performance.

Neo4j

Neo4j® on IBM POWER8 is the result of a joint effort between Neo4j and IBM engineering to provide a scalable graph database platform capable of storing and processing graphs of extremely large size and removing previous real-time scalability limits.

Cassandra

IBM has released enhacements to open source version of Cassandra to take advantage of IBM Data Engine for NoSQL platform capability. This improves througput performance of read-intensive workloads.