Introduction to IBM PureFlex System and cloud
Distributed computing has undergone a transformation over the past decade, with the introduction and mass adoption of hypervisor-based virtualization technology. As this technology has evolved, it has driven a re-architecture of traditional data center infrastructure layouts, enabling new approaches to workload placement.
Originally, deploying virtualization simply enabled multiple discrete workloads to be placed on the same physical server, leading to higher asset utilization and server consolidation. However, as this technology has matured, coupled with new industry standards driving new features, and higher virtual machine (VM) density per physical server was achieved, it has also begun to impact traditional interactions between server, storage, and network entities, including entity management and administration.
The industry was looking for a new approach. Hence, the concept of converged infrastructure, or expert integrated systems was born. Expert integrated systems package traditional IT resources into a single optimized solution, with prepacked components including servers, storage devices, networking equipment, and software for IT infrastructure management, automation, and orchestration. With this evolution of technology, we moved from discrete, siloed, and underutilized IT resources to shared resource pools. These resource pools, enabled by expert integrated systems, exploit inherent features of advanced hypervisor-based virtualization and cloud computing. To meet these new industry requirements, IBM developed a completely new integrated data center solution.
In 2012, IBM announced IBM PureSystems, an offering based on preconfigured software, servers, and storage that form an expert integrated system.
1.1 Introduction to IBM Flex System hardware
IBM Flex System hardware provides the building blocks for IBM PureFlex System and IBM PureApplication™ System.
IBM Flex System is at the heart of the new type of data center building blocks. The IBM Flex System can meet the growing needs of any organization that requires the most advanced blade system to date. The design of the Flex System hardware is focused on long term investments, performance, and flexibility. This focus allows Flex System to support client needs now and with future generations of technology.
IBM Flex System is based on four main integrated components:
Compute nodes
Storage
Networking
Management
IBM Flex System compute nodes support multiple compute architectures and up to four separate operating systems. The converged infrastructure and advanced management aim to improve utilization and productivity to ultimately lower the cost of delivering IT services. The storage capabilities are built for performance and scalability.
IBM PureFlex System offering is designed for private clouds. It is available in three configurations:
Express
Standard
Enterprise
By combining the flexibility of the general purpose PureFlex System with SmartCloud Entry, clients can add cloud functionality to their converged environment. SmartCloud Entry provides enterprises with the capability of managing their multiple architecture cloud solution on a single system.
1.1.1 Three configurations
PureFlex combines the components of IBM Flex System hardware into expert integrated systems. They are factory-preconfigured in Express, Standard, and Enterprise configurations to meet a wide variety of needs. All of the PureFlex configurations include a PureFlex Enterprise Chassis incorporated into a PureFlex 42U Rack, and also a standard IBM Flex System Manager™ (FSM) as described in section 1.5. Each configuration brings a different level of power and cooling, appropriate to the size of the intended deployment.
IBM PureFlex System Express
The IBM PureFlex System Express configuration is the entry point in the PureFlex family. It includes, as standard, two power supplies (out of a maximum of six), and two fans (maximum of eight).
IBM PureFlex System Standard
The PureFlex Standard configuration adds two more power supplies, for a total of four.
IBM PureFlex System Enterprise
The PureFlex Enterprise configuration is preconfigured with the maximum of six power supplies allowed in an Enterprise chassis, and also all eight configurable fans. In this configuration, the maximum number of PureFlex compute nodes can be supported with N+N power supply redundancy, providing the highest levels of compute power and redundancy.
1.2 IBM Flex System x86 Nodes
IBM Flex System x86 nodes are built on Intel Xeon processors and support a wide range of virtualization technologies and operating systems. There are several members of the IBM Flex System x86 family. Only the IBM Flex System x240 systems were used in the sample environment for this writing.
1.2.1 IBM Flex System x240
The IBM Flex System x240 is a two-socket, single-bay Flex node based on Intel Xeon processors. The system supports up to 768 GB of RAM and up to 2 TB of internal storage. Integrated 10GbE networking and system management make the x240 an excellent choice for a wide variety of workloads, including mainstream virtualization.
1.3 IBM Flex Power Systems nodes
IBM Flex Power Systems™ nodes bring the Power architecture to the Flex family. The Flex Power Systems nodes are built on IBM POWER7® and IBM POWER7+™ technology, which bring faster frequencies and larger L3 cache sizes and improve performance on most workloads. POWER7 family processors are also extremely efficient and provide more performance per watt of energy consumed. Both Flex Power Systems nodes, described next, support IBM AIX®, IBM i, and Linux.
1.3.1 IBM Flex System p260 compute node
The Flex System p260 compute node is a single-wide node and is available with 8 or 16 cores in speeds in the range of 3.2 - 4.1 GHz. The p260 supports up to 512 GB of RAM.
1.3.2 IBM Flex System p460 compute node
The Flex System p460 compute node provides four sockets in a double-wide form-factor. This allows a p460 to have either 16 or 32 POWER7 cores, with 4 MB of L3 cache per processor core. With a maximum of 1 TB of RAM per node, the p460 is suitable for memory and compute-intensive workloads, and also for meeting the demands of heavy virtualization.
1.4 IBM Flex System V7000
The IBM Flex System V7000 is the IBM Storwize® V7000 adapted to the IBM Flex System ecosystem. It can house up to 240 drives for each control enclosure. Currently supported drives include solid-state drives (SSDs), nearline SAS, 10,000 RPM (10k) SAS, and 15k SAS drives. Built-in support for tiering and advanced replication features make the most efficient use of the installed drives and allow a variety of configurations to suit many environments.
1.5 IBM Flex System Manager
The IBM Flex System Manager (FSM) is the center of IBM Flex System management. The FSM provides a real-time interactive chassis map to simplify monitoring and management of resources in the Enterprise chassis. The FSM provides access to power and health monitoring of nodes and also configuration of installed nodes, and simplifies deployment of workloads across compute nodes. It is a central point of management for every element in the Enterprise chassis.
A single FSM can manage up to four Enterprise chassis.
1.6 IBM Flex System Enterprise Chassis
The IBM Flex System consists of one or more rack-mounted chassis. Each chassis can contain two Chassis Management Modules (CMM), up to 14 compute nodes, six power supplies, and 10 fans.
1.6.1 Chassis Management Module (CMM)
The CMM is a single-chassis manager that communicates with the individual compute note management controllers. CMM supports system monitoring, event recording, and alerts. It also provides an interface for the management of the chassis, its devices, and the compute nodes. Each chassis supports two CMMs for higher availability.
1.6.2 Midplane
The midplane in an IBM Flex System Enterprise chassis serves several functions:
It is responsible for distributing power. It does this in a single power domain, distributing power to the compute nodes, the IO modules (scalable switch elements), and ancillary components.
It provides the physical connectivity from network and storage adapters in each node to corresponding ports in the IO modules.
It contains apertures which connect the cooling channels in the front of the enterprise chassis to the appropriate channels in the rear of the chassis to provide better airflow and more efficient cooling.
The IBM Flex System Enterprise chassis midplane design differs from the backplane design used in many blade chassis in that it eliminates active components. The midplane is designed for high-frequencies and will support the next generation of networking and storage signalling protocols.
1.6.3 Power
Power in the Flex System Enterprise chassis is provided by up to six power supplies that can supply 2500W each. This flexibility allows you to grow your power capacity according to your needs. The power supplies in an Enterprise chassis can be configured for N+1 or N+N redundancy to provide the combination of power and availability most appropriate for the target environment.
1.6.4 Cooling
The Enterprise chassis is designed to provide excellent cooling capacity to the installed nodes, I/O modules and management modules. Cooling is divided into two zones, each of which allows for up to four 80 mm fans.
1.7 IBM Flex System scalable switch elements
Scalable switch elements provide the Flex System nodes and the enterprise chassis with access to networking and storage resources. The Enterprise chassis provides space for up to four scalable switch elements. These switch elements can be any combination of network, storage, or converged IO devices. Several options are available. Following is a description of the modules employed for the purposes of this publication.
1.7.1 IBM Flex System Fabric EN4093 10 Gb Scalable Switch
The EN4093 is a 10 Gb Layer 2/3-managed Ethernet switch that provides up to 42 internal and 22 external ports. The basic configuration of the EN4093 provides 14 internal 10GbE ports and 10 external 10GbE ports. Two Features on Demand upgrades are available. With upgrade 1 applied, the switch provides 28 internal 10GbE ports, 10 external 10GbE ports, and two external 40GbE ports. With upgrade 2 applied, the switch provides 42 internal 10GbE ports, 14 external 10GbE ports, and four external 40GbE ports. This, as with several other features of the IBM Flex System products, allows the switch to grow to meet the demands of the data center without wasting capacity that is not needed. Support for advanced virtualization features such as Virtual Fabric and IBM VMready® make the switch well suited to highly virtualized environments.
1.7.2 IBM Flex System FC3171 8Gb SAN Switch
The FC3171 is an 8 Gb SAN switch that provides 14 full-duplex, internal ports and six external ports. The FC3171 is available as a pass-through switch, to expose only the host ports to the SAN, and as a regular SAN switch.
 
..................Content has been hidden....................

You can't read the all page of ebook, please click here login for view all page.
Reset