IBM Power Systems Update 1Q17
Download as PPTX, PDF1 like858 views
This document provides a summary of an IBM Power Systems product update presentation. It discusses how organizations struggle with challenges around understanding data, keeping pace with innovation, and accelerating time-to-market. It then provides examples of how IBM Power Systems and hybrid cloud solutions can help organizations overcome these challenges by enabling insights from large amounts of data, superior performance, and faster delivery of products and services. The document also includes information about the Power8 architecture and its performance advantages over Intel systems.
IBM Power Systems Update 1Q17
- 1. © IBM Corporation, 2016 IBM Power Systems Update 19th January 2017 Presented by David Spurway IBM Power Systems Product Manager IBM Systems, UK and Ireland
- 2. Organizations struggle with key technology challenges supporting insights, change, and speed Understanding Data Digital intelligence is a primary competitive advantage Yet 88% of all data today is unstructured and invisible to computers1 “The next wave will be all about connecting dots and correlating data to produce actionable insights.” –CIO, Retail, United States2 Keeping Pace Superior technology delivers superior performance Yet the pace of innovation (72%) is now the top CEO challenge, even greater than security (66%)3 “Disruptive technologies could change the fundamentals of our business” –Kazuo Hirai, CEO, Sony Corporation, Japan4 Accelerating Time-to- Market Sustainable success is now measured in days, not weeks Outperforming CxOs are 95% more likely to focus on being first to market4 “We’ve been charged with speeding up time-to-market, both for the products we sell and for our own internal tools.” ⎯Kalev Reiljan, CIO, TeliaSonera, Finland2 Source: 1) IBM Research Source: 2) IBM Institute for Business Value, Redefining Connections, Insights from the Global C-suite Study – The CIO Perspective Source: 3) Fortune, Myth-busting the Fortune 500, 2015 Source: 4) IBM Institute for Business Value, Redefining Boundaries: Insights from the Global C-Suite Study, 2015 IBM Systems | 2
- 3. Expose systems as APIs to enable composable services IBM Systems | 3 Architects of the future require IT infrastructure that can do more than ‘just work’ Servers and storage are no longer inanimate. They can understand, reason, and learn. Today, they can think. Outthink status quo. Think IT infrastructure for the cognitive era. Detect anomalies to proactively resolve issues Move data to right location based on usage patterns Deliver real-time insights from oceans of data
- 4. 4 © IBM Corporation, 2016 A surprise trip to Lisbon An Insurance Story https://www.linkedin.com/pulse/surprise- trip-lisbon-insurance-story-david-spurway
- 5. 5 © IBM Corporation, 2016 I find myself in Lisbon… • My Parents-In-Law set off on a cruise, but Mother-In-Law is taken ill • Hospitalised in Lisbon • Insurance company take 12 days to validate cover • Father-In-Law is the real concern • Twitter storm
- 6. 6 © IBM Corporation, 2016 Insurance industry - Capturing hearts, minds and market share How connected insurers are improving customer retention http://www-935.ibm.com/services/us/gbs/thoughtleadership/insuranceretention/
- 7. 7 © IBM Corporation, 2016 Insurance is a product based on trust, for which perception matters.
- 8. 8 © IBM Corporation, 2016 Mix in some Bluemix and Watson…
- 9. 9 © IBM Corporation, 2016 Unstructured Facebook entries in, Geo Political Entities out! http://ibmlaser.mybluemix.net/siredemo.html
- 10. 10 © IBM Corporation, 2016 Who & Where: Find out with Bluemix Geospatial Analytics https://developer.ibm.com/bluemix/2014/12/17/find-bluemix-geospatial-analytics/
- 11. 11 © IBM Corporation, 2016 But what does this have to do with IBM Power? • Details of customers are on IBM Power Systems –Customer Details –Declared Medical History • Location, Sentiment, etc. are on Social Media • Other data could be available through B2B APIs • Hybrid Cloud solution could differentiate from competition
- 12. 13 © IBM Corporation, 2016 POWER8 Overview Optimized for Data Open Innovation Platform Superior Cloud Economics
- 13. 14 © IBM Corporation, 2016 Off-premise On-premise VPN SecurityServices SecurityServices Container Systems of Record Systems of Engagement Bluemix AIX RHEL IBM i IBM Systems Hybrid Cloud Reference Architecture
- 14. 15 © IBM Corporation, 2016 IBM Power Systems Enterprise Cloud Infrastructure On-Premises Cloud Hybrid Infrastructure Complementary Built-in Cloud Deployment Service Options Transform traditional infrastructure with automation, self-service and elastic consumption models Securely extend to Public Cloud with rapid access to compute services and API integration • OpenStack-based Cloud Management: enabling DevOps to Full production • Open source automation (installation and config. recipes) • Flexible elastic private cloud capacity and consumption models • Cross Data Center Inventory and Performance Monitoring via the IBM Cloud • Manage VMs across on and off-premises clouds with a single pane of glass (e.g., VMware vRealize) • Securely connect traditional workloads with cloud- native apps (Power & API Connect, BlueMix) • Optional DR as a Service (GDR for Power) • Free access and capacity flexibility with SoftLayer - Free SoftLayer starter pack (12 server months) - Flexibility to run capacity On Premises or in SoftLayer • Design for Cloud Provisioning and Automation • Build for Infrastructure as a Service • Build for Cloud Capacity Pools across Data Centers • Design for Hybrid Cloud with BlueMix • Deliver with automation for DevOps • Deliver with Database as a Service
- 15. 16 © IBM Corporation, 2016 CCI (VMs) Off-premise On-premise AIX X LINUX IBM i VPN Intel SecurityServices SecurityServices Nova PowerKVM PowerVM 2.5 1.3 Novalink HMC Pre-POWER8 Nova Partition OpenStack Services Nova API Nova Core POWER8 Power & Hybrid Cloud Architecture for IaaS with PowerVC & IBM Cloud Orchestrator POWER8 PowerVM LINUX Baremetal Intel PowerKVM Nov a LINUX A I X IBM i
- 16. 17 © IBM Corporation, 2016 Off-premiseOn-premise AIX X LINUX IBM i VPN Baremetal Intel SecurityServices SecurityServices IBM PowerVC Cloud Edition (openstack liberty) PowerVM Power & Hybrid Cloud Architecture for IaaS with Openstack Novalink Nova Partition OpenStack Services Nova API Nova Core PowerKVM Self Service Catalog Metering Multitenancy PowerVM HMC Nova PowerKVM LINUX AIX X IBM i LINUX Pre-POWER8 POWER8 POWER8
- 17. 18 © IBM Corporation, 2016 Off-premiseOn-premise AIX RHE L IBM i Power & Hybrid Cloud Architecture for PaaS VPN Baremetal Intel SecurityServices SecurityServices Patterns Creation & Deployment Middleware& Software Service Middleware & Software Engine Patterns Designer Service Patterns Engine UrbanCode Deploy Power KVM CCI (VMs) Intel Novalink Nova Partition OpenStack Services Nova API Nova Core Power KVM
- 18. 20 © IBM Corporation, 2016 POWER8 Overview Optimized for Data Open Innovation Platform Superior Cloud Economics
- 19. 21 © IBM Corporation, 2016 Chip Bandwidth and Threading 2S=2 Socket, 4S=4 Socket, 8S=8 Socket, DCM=Dual Chip Module, SCM=Single Chip Module, Low Power Models Not Included, Updated 6/1/2015 Chip Family Memory Bandwidth per Socket Peak I/O Bandwidth per Socket Threads per Core Intel 26xx-V3 (2S, 8+ Cores) 59 – 68 GB/s 80 GB/s 1, 2 Intel 48xx-V3 (4S, 8+ Cores) 102 GB/s 64 GB/s 1, 2 Intel 88xx-V3 (8S, 8+ Cores) 102 GB/s 64 GB/s 1, 2 POWER8 (DCM, 8+ Cores) 192 GB/s 96 GB/s 1, 2, 4, 8 POWER8 (SCM, 8+ Cores) 230 GB/s 64 GB/s 1, 2, 4, 8
- 20. 22 © IBM Corporation, 2016 Single Thread Performance is Increasing Performance measured in rperfs +14 % +28% POWER7 740 3.7 GHz 8 Cores SMT1 56 POWER7+ 740 4.2 GHz 8 Cores SMT1 64 POWER8 S824 4.1 GHz 8 Cores SMT1 82
- 21. 23 © IBM Corporation, 2016 POWER8 vs. Haswell Performance Benchmark Intel E5 V3 Haswell Best Results POWER8 Best Results Per Core Ratio SAP SD 2-Tier ERP 6 Dell PowerEdge R730 E5-2699 v3, 36 Core 16,500 IBM E870 80 Core 79,750 2.2 X SPECjbb2013 Lenovo Flex x240 E5-2699 v3, 36 Core 245,178 IBM E870 80 Core 1,299,150 2.4 X SPECint_rate2006 Dell PowerEdge T620 E5-2699 v3, 36 Core 1,400 IBM E880 64 Core 5,400 2.1 X SPECfp_rate2006 Dell PowerEdge T620 E5-2699 v3, 36 Core 942 IBM E880 64 Core 4,470 2.5 X Oracle e-BS 12.1.3 Extra Large Payroll Cisco UCS C240 M4 E5-2697 v3, 28 Core 1,125,281 IBM S824 12 Core 1,090,909 2.3 X IBM Power E870 on the two-tier SAP SD standard application benchmark running SAP enhancement package 5 for the SAP ERP 6.0 application; 8 processors / 80 cores / 640 threads, POWER8; 4.19GHz, 2048 GB memory, 79,750 SD benchmark users, running AIX® 7.1 and DB2® 10.5, dialog response: 0.97 seconds, order line items/hour: 8,722,000, dialog steps/hour: 26,166,000, SAPS: 436,100, Database response time (dialog/update): 0.013 sec / 0.026 sec, CPU utilization: 99%, Cert #2014034. Result valid as of November 15, 2015. Source: http://www.sap.com/benchmark . Dell PowerEdge R730, on the two-tier SAP SD standard application benchmark running SAP enhancement package 5 for the SAP ERP 6.0 application; 2 processors/36 cores/72 threads, Intel Xeon Processor E5-2699v3; 2.30 GHz, 256 GB memory; 16,500 SD benchmark users, running RHEL 7 and SAP ASE 16; Certification # 2014033. SPECjbb2013 results are submitted as of 12/01/2014. For more information go to http://www.specbench.org/jbb2013/results SPECcpu2006 results are submitted as of 11/15/2015. For more information go to http://www.specbench.org/cpu2006/results All results use Oracle eBS 12.1.3 Payroll Batch Extra Large Kit and are current as of 11/15/2015. For more information go to http://www.oracle.com/us/solutions/benchmark/apps-benchmark/results-166922.html
- 22. 24 © IBM Corporation, 2016 Power S814 Power S824 Power S822 Power S812L Power S822L Scale-out Systems (1 & 2 sockets) IBMPowerSystems Enterprise Systems (4+ sockets) Power E880CPower E870C Power E850C Power S824L Power Systems Range Operating Systems or Hypervisors Management
- 23. 25 © IBM Corporation, 2016 AIX 7.2 Server Based Flash Caching AIX Flash Cache Engine Workload Flash / SSD AIX LPAR SAN LUN LUN LUN I/O APIs (File System, Device, etc.) • Server caches SAN data in server attached storage (SSDs, SAS attached drawer, FC connected flash, etc.) • Cache is read-only to service application reads. Writes go straight through to the SAN back-end. • Potential benefits are performance (more transactions, reduced response time), and/or improved scale and efficiency of SAN infrastructure via SAN off-load • For workloads that use RAM as a cache for I/O, smaller RAM caches maybe be possible with a flash based cache • Provides storage vendor agnostic I/O caching solution integrated into AIX • Cache is transparent to workloads and requires no application code changes • Supports virtualization features such as Live Partition Mobility • LPM support requires VIOS 2.2.4 (Dec 2015) • For more Information - https://ibm.biz/AIX72_Flash_Cache IBM, EMC, Hitachi, etc. Also available on AIX 7.1 TL4/SP2 (May 2016)
- 24. 26 © IBM Corporation, 2016 SAN cachedisk0 cachedisk2 Cache Pool (SSD disks) hdisk1 hdisk2 hdisk3 hdisk8 SSD LPAR1 Virtualized: Flash LUN managed by VIOS Cache management VIOS LPAR2 Cache Engine Cache Engine Cache partition 1 Cache partition 2 LPM Capable AIX 7.2 Flash Cache Configurations * Requires VIOS 2.2.4 (Dec 2015) Dedicated: Dedicated flash LUN SAN Cache Pool (SSD disks) hdisk1 hdisk2 hdisk3 hdisk8 Cache Partition SSD LPAR CM / Cache Engine
- 25. 27 © IBM Corporation, 2016 OLTP with 8 cores – 10 users Flash Caching Example AIX Flash Cache Engine Workload Flash / SSD AIX LPAR SAN LUN LUN LUN I/O APIs (File System, Device, ..) IBM, EMC, Hitachi, etc. IBM, EMC, Hitachi, etc. Flash caching 33% of database provides 77% TPM improvement IBM Montpellier Customer Benchmarking Center
- 26. 28 © IBM Corporation, 2016 AIX 7.2 Next Generation AIX Live Update Technology • Project Requirements –Use existing maintenance model and administrative tools (NIM, geninstall) –Start with live AIX kernel replacement for i-fixes, then in future extend to SPs and TLs –Long term goal is to provide non-disruptive updates for all of AIX (kernel, commands, and Libs) Original rootvg New rootvg (Alt-disk-install) VIOS NPIV / VSCSI SEA Original LPAR Virt. ent VSCSI Surrogate LPAR Virt. ent CEC Original rootvg mirror New rootvg (Alt-disk-install) Original Kernel Updated Kernel VSCSI Running Workload Running Workload Live Migration
- 27. 29 © IBM Corporation, 2016 IBM AIX 7.2 offers new enhancements for reduced downtime, acceleration, and efficient storage utilization • https://www- 01.ibm.com/common/ssi/ShowDoc.wss?docURL=/common/ssi/rep_ca/5/877/ ENUSZP16-0525/index.html&lang=en&request_locale=en –October 11, 2016 –#AIX 7.2 enhancements inc non-disruptive Live Kernal Update for SPs and TLs! Also LVM thin storage block reclamation
- 28. 31 © IBM Corporation, 2016 POWER8 Overview Optimized for Data Open Innovation Platform Superior Cloud Economics
- 29. 32 © IBM Corporation, 2016 OpenPOWER drives industry innovation The OpenPOWER Foundation creates an open ecosystem, using the POWER Architecture to share expertise, investment, and server-class intellectual property to serve the evolving needs of customers. Performance of leading POWER architecture Broadens the capability and performance of the POWER platform Collaboration across multiple thought leaders Collaborative development model drives collective thought leadership, simultaneously across multiple disciplines Open Development OpenPOWER enables greater innovation through both open software and open hardware
- 30. 33 © IBM Corporation, 2016 OpenPOWER Open Interfaces OpenPOWER open interfaces enable an unbeatable innovation pace CAPI NVLink 40 GB/s CAPI 16 GB/s POWER8 Memory Interface Control Server Class Memory DMI IBM and Partner Devices GPU
- 31. 34 © IBM Corporation, 2016 What is CAPI? FPGA POWER8 Core PCIe POWER8 Processor OS App Memory (Coherent) AFU IBM Supplied PSL Virtual Memory CAPP CAPI (Coherent Accelerator Processor Interface) is a set of IBM innovations in hardware and software that allow an application and it’s accelerated component to share the same virtual address space. For customers, CAPI enables performance and ease of programming: • Application to set up data coherently in shared virtual memory and call the accelerator functional unit (AFU) • AFU to reads and writes data from and to shared virtual memory coherently across PCIe and communicating with the application • AFU to coherently cache data in the PSL cache for quick AFU access
- 32. 35 © IBM Corporation, 2016 Power Systems and NVIDIA GPU NVIDIA GPU NVIDIA GPU with NVLink 2015 Power Chip Power Chip with NVLink 2016 80 GB/s Peak* PCIe x16 32 GB/s
- 33. 36 © IBM Corporation, 2016 Hadoop / Spark Commercial Technical OpenPOWER Linux Cluster (LC) Systems (existing) S812LC 1 socket, 2U, Linux 8 or 10 cores Up to 1 TB memory Up to 112 TB Storage 4 Available PCI Slots KVM / Bare Metal S822LC – GCA 2 socket, 2U, Linux 16 or 20 cores Up to 1 TB memory 2 Disks 5 Available PCI slots KVM / Bare Metal S822LC – GTA 2 socket, 2U, Linux 16 or 20 cores Up to 1 TB memory 2 Disks 2 NVIDIA K80 GPUs 3 Available PCI Slots Bare Metal
- 34. 37 © IBM Corporation, 2016 Dense Virtualisation Big Data Technical OpenPOWER Linux Cluster (LC) Systems (new) S821LC Up to 2 socket, 1U, Linux 8, 10, 16 or 20 cores Up to 512 GB memory Up to 38 TB Storage 1 NVIDIA K80 GPU 4 Available PCI Slots S822LC for Big Data Up to 2 socket, 2U, Linux 8, 10, 16 or 20 cores Up to 512 GB memory Up to 96 TB Storage 2 NVIDIA K80 GPUs 5 Available PCI slots S822LC for HPC 2 socket, 2U, Linux NVLink 16 or 20 cores Up to 1 TB memory 2 Disks 2 or 4 NVIDIA P100 GPUs 3 Available PCI Slots
- 35. 38 © IBM Corporation, 2016 OpenPOWER Linux Cluster (LC) Systems S821LC MTM: 8001-12C Code Name: Stratton S822LC for Big Data MTM: 8001-22C Code Name: Briggs S822LC for HPC MTM: 8335-GTB Code Name: Minsky (was Garrison) Power S812LC MTM: 8348-21C Code Name: Habanero Power S822LC MTM: 8335-GCA Code Name: Firestone Power S822LC MTM: 8335-GTA Code Name: Firestone
- 36. 39 © IBM Corporation, 2016 Augmented intelligence, Artificial Intelligence, Cognitive
- 37. 40 © IBM Corporation, 2016
- 38. 41 © IBM Corporation, 2016 Introducing PowerAI: Get Started Fast with Deep Learning Package of Pre-Compiled Major Deep Learning Frameworks Package of Pre-Compiled Major Deep Learning Frameworks Package of Pre-Compiled Major Deep Learning Frameworks Enabled by High Performance Computing Infrastructure
- 39. 42 © IBM Corporation, 2016 PowerAI takes advantage of NVLink between POWER8 & P100 to increase system bandwidth • NVLink between CPUs and GPUs enables fast memory access to large data sets in system memory • Two NVLink connections between each GPU and CPU-GPU leads to faster data exchange P100 GPU POWER8 CPU GPU Memory System Memory P100 GPU 80 GB/s GPU Memory NVLink 115 GB/s P100 GPU POWER8 CPU GPU Memory System Memory P100 GPU 80 GB/s GPU Memory NVLink 115 GB/s
- 40. 43 © IBM Corporation, 2016 Google, Rackspace, and GPUs: OH MY! See what you missed at OpenPOWER Summit http://openpowerfoundation.org/blogs/google-rackspace-gpus-openpower-summit/
- 41. 44 © IBM Corporation, 2016http://www.nextplatform.com/2016/08/24/big-blue-aims-sky-power9/
- 42. 45 © IBM Corporation, 2016http://www.nextplatform.com/2016/08/24/big-blue-aims-sky-power9/
- 43. © IBM Corporation, 2016 Questions? David Spurway – IBM Power Systems Product Manager Email: david.spurway@uk.ibm.com Phone: 07717 892 896 Twitter, LinkedIn, YouTube
- 44. 47 © IBM Corporation, 2016
- 45. © IBM Corporation, 2016 Thank you! David Spurway – IBM Power Systems Product Manager Email: david.spurway@uk.ibm.com Phone: 07717 892 896 Twitter, LinkedIn, YouTube