This chapter discusses how IBM POWER8 can help clients with their analytic workload requirements, and it describes the advantages that this solution delivers to help clients manage and understand data to gain competitive advantages.

This chapter contains the following topics:

A high-performance infrastructure that is scalable for varying workloads, highly available, and optimized for price performance is critical for getting the most out of the analytics platform and deriving timely insight. It helps infuse analytics seamlessly to drive better business outcomes and allows companies to build on what they already have while they add new capabilities as their needs grow.

IBM Solution for Analytics - Power Systems Edition delivers a solution for business intelligence and predictive analytics. It is flexible and integrated, and it provides options to pre-load and configure one or more IBM analytics applications with data warehouse acceleration on a POWER8 processor-based server.

These analytics tools, with many other tools in the IBM analytics suite, help turn big data into actionable insights to help companies address customer retention and growth, IT costs, management of security risks, counter fraud techniques, optimization of IT operations, and digital innovation opportunities.

IBM Power Systems help clients accelerate results for analytics applications. Imagine the possibilities that fast, efficient analytics can open for you. How many more questions can you ask if you can get answers immediately? To improve your business’ time to insight, you need an analytics infrastructure that is easy to set up, that is flexible and scalable, and that offers powerful analytics applications.

Helping you make the best decision for your business means understanding what is happening, why it is happening, what can happen in the future, and what you need to do fast and efficiently. Therefore, having the data processing capacity, accelerating the knowledge with this infrastructure, and helping you to deliver faster business outcomes with predictive analytics can provide you with the solution to meet your constant business needs.

To provide an infrastructure for big data and analytics, the IBM hardware and software teams work together to optimize the entire solution stack for data and analytics. The suite of capabilities that are built on IBM POWER8 can help you develop a thorough plan for addressing your big data and analytics needs.

By integrating data environments across your business, building on a proven infrastructure as IBM Power Systems, using storage solutions that maximize data efficiency, and taking advantage of the latest and best data management and analytics tools, you can accelerate your insights while you reduce costs. As a foundation of the IBM comprehensive big data and analytics portfolio, Power Systems continues to be a key driver in pushing your business forward in the race toward digital transformation with predictive analytics.

With an infrastructure, such as the IBM POWER8 that is tuned for big data and analytics, you can build analytics into operational processes. This POWER8 infrastructure can scale effortlessly while it enables highly secure, shared access, resiliency of the information, and generate insights that are needed to make better decisions.

The POWER8 infrastructure helps to provide continuous availability to address more users, more concurrency, changing demand, and resiliency of the data.

For example, the IBM Data Engine for Analytics provides a customizable, pre-integrated infrastructure solution that brings together the innovative capabilities of IBM Power Systems (massively parallel compute capacity for a fast time to get the result), IBM Elastic Storage Server (software defined storage for fast access to petabytes of data), and IBM Platform Computing (intelligent cluster automation and management for ease of deployment and operations).

This solution can be combined for faster time to value with big data and analytics software, such as IBM InfoSphere BigInsights for Hadoop for analytics workloads, IBM InfoSphere Streams for high ingest streaming analytics, and IBM Watson™ Explorer analytical components for advanced natural language processing of unstructured data.

The suite of capabilities that are built on IBM POWER8 can help businesses develop a thorough plan for addressing their big data and analytics needs. IBM offers a broad set of analytics capabilities built on the proven foundation of a single platform, IBM Power Systems. Power Systems is an open, secure, and flexible platform that is designed for big data. It has massive input/output (I/O) bandwidth to deliver analytics in real time, and it can provide the capabilities that are needed to handle the varying analytics initiatives that each business requires.

Transactional memory (TM) is a shared-memory synchronization construct that allows process-threads to perform sequences of storage operations that appear to be atomic to other process-threads and applications. This capability allows optimistic execution as a means to take advantage of the inherent parallelism that is found in the latest generation of Power Systems.

One of the main uses of TM is the speed-up of lock-based programs by using the speculative execution of lock-based critical sections (CSs) without first acquiring a lock. This function allows applications that were not carefully tuned for performance to take advantage of the benefits of fine-grain locking. The transactional programming model also provides productivity gains when you develop lock-based shared memory programs.

Applications can also use TM to checkpoint and restore the architectural state, independently of the atomic storage access guarantees that are provided by TM.

The POWER8 processor benefits from the next generation Peripheral Component Interconnect Express (PCIe) Gen3 and shorter hardware paths to the adapters. The POWER8 chip design includes on-chip PCIe buses, which means fewer hardware delays and lower latency to PCIe slots. This design results in lower latency for networking and storage protocols, reducing latency by over 1 µs from the previous I/O hub-based POWER7 Systems.

The PCIe Gen3 x16 bus increases the PCIe bus peak bandwidth to 112 Gbps (for a single x16 bus), which is about four times the bandwidth in previous POWER PCIe Gen2 x8 slots in POWER7 and IBM POWER7+™ processor-based systems. Each processor module or socket provides two or four PCIe buses, depending on the system model. Several buses are x16 and others are x8. Depending on the system model, several buses can connect to PCIe I/O hub chips on the system board.

These PCIe slots are for slower-speed adapters so they can share a single higher-speed bus. In addition, for systems that support I/O drawers, the PCIe buses in the I/O drawer all use a PCIe I/O hub. For applications that are sensitive to latency or those applications that require high bandwidth or high message rates, the adapters that are used by the application need to use, when possible, the PCIe slots in the chip where the application runs.

In all cases, bandwidth must be the same no matter where in a POWER8 processor-based system an adapter is plugged. However, for latency, small increases are possible if the adapter is in a different socket, node, or central processor complex drawer.

As a rule, high-speed and low-latency adapters must be placed in the direct PCIe slots on each socket. The PCIe slots behind I/O hubs and in I/O drawers must be used for lower bandwidth and non-latency sensitive adapters. Many adapters are restricted to certain slots. Consult a Power Systems PCI Adapter Placement Guide for a specific adapter.

The POWER8 processor contains many new and important performance features, such as support for eight hardware threads in each core and support for transactional memory. The POWER8 processor is a strict superset of the IBM POWER7+ processor, and so all of the performance features of the POWER7+ processor, such as multiple page sizes, also appear in the POWER8 processor.

For more information about POWER performance, see the publicationPerformance Optimization and Tuning Techniques for IBM Power Systems Processors Including IBM POWER8, SG24-8171, at the following website:

IBM Spectrum Scale is software-defined storage for high-performance, large-scale workloads on-premises or in the cloud. Built on the award-winning IBM General Parallel File System (GPFS), this scale-out storage solution provides file, object, and integrated data analytics for the following areas:

IBM Data Engine for Analytics is a customizable, pre-integrated infrastructure solution that brings together the innovative capabilities of IBM Power Systems (massively parallel compute capacity for fast time to result), IBM Elastic Storage Server (software defined storage for fast access to petabytes of data), and IBM Platform Computing (intelligent cluster automation and management for ease of deployment and operations).

For faster time to value, big data and analytics software, such as IBM Open Platform with Apache Hadoop, IBM InfoSphere Streams for high ingest streaming analytics, and IBM Watson Explorer analytical components for advanced natural language processing of unstructured data, can be preinstalled and configured before shipment to the client site.

For additional information about IBM Spectrum Scale and the feature that gives analytics implementations a push, see 2.3.5, “Spectrum Scale and File Placement Optimizer” on page 33.

For more information about IBM Spectrum Scale, see the IBM Spectrum Scale (formerly GPFS), SG24-8254, at the following website: