Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

Free Your Analytics From Paralysis – Infrastructure Matters

234 views

Published on

Exploring new data sources, such as social media, mobile devices and the Internet of Things, is a gold mine for business analytics to discover valuable information for business decisions. The dark side: a dramatically greater amount of data needs to be stored, moved and processed in order to uncover the secrets hidden in floods of Big Data. Only a powerful and optimized data center infrastructure will enable you to extract, collect and analyze data in the best manner and to provide the high performance required to get the desired results. If you wish to learn more, then you should not miss this session.
Speaker:
Gernot Feis

Published in: Technology
  • Be the first to comment

  • Be the first to like this

Free Your Analytics From Paralysis – Infrastructure Matters

  1. 1. 0 Copyright 2016 FUJITSU Fujitsu Forum 2016 #FujitsuForum
  2. 2. 1 Copyright 2016 FUJITSU Free Your Analytics From Paralysis – Infrastructure Matters Gernot Fels Head of Integrated Systems, Global Product Marketing, Fujitsu
  3. 3. 2 Copyright 2016 FUJITSU Digitalization resounds throughout the land  Key benefits of digitalization  Business responsiveness  Customer retention and loyalty  Workforce productivity  Attract and retain talent  Digitalization is based on (big) data and its analytics  Know things you don’t know  Predict opportunities  Minimize risks  Take better and faster decisions Digitalization won’t work without data analytics.
  4. 4. 3 Copyright 2016 FUJITSU Big Data – what, why, how … Volume Variety Versatility Velocity Value New value by correlating data and new ways of analytics.
  5. 5. 4 Copyright 2016 FUJITSU Examples: What customers achieved …  Customer  Debt collection agency  Challenge  Identify 1.5M individuals not listed in payment plan  £2 billion of uncollected debt  Data analytics helped  Understand debtors (e.g. behavior, propensity to pay)  Create right payment plans  Outcome  Collect £12.7M within 9 months  Customer  Financial services provider  Challenge  How to retain more clients  Data analytics helped  Identify customer behavior  Develop effective contact strategy with correct up-sell message  Outcome  Increase customer value by 74%  Reduce churn by 17% Making analytics pay Understanding each and every customer
  6. 6. How to get from Big Data to Big Value? A lot of questions …  Which use case?  Which business outcome?  Which data from which sources?  How to transform data into high quality?  What to look for?  Which questions to ask?  Which actions to trigger?  Which analytic methods?  How to visualize results effectively?  Which tools?  How to ensure data security and privacy?  What about data retention and deletion?  … Creative ideas, deep knowledge of data, tools and intended results.
  7. 7. 6 Copyright 2016 FUJITSU Big Data Infrastructure Matters – If You Want to Avoid Paralysis “There are organizations that feel paralyzed in their business operations. Analytics can help, but without the right infrastructure, paralysis will not go.”
  8. 8. 7 Copyright 2016 FUJITSU Big Data infrastructure: Challenges Big Data changes the infrastructure conversation. Exponential growth of data  Keep processing time constant while volumes increase  Store and process large data volumes at affordable cost  Ad-hoc queries and real-time demands  Process continuously generated event streams
  9. 9. 8 Copyright 2016 FUJITSU Big Data infrastructure master plan Data Sources Analytics Platform Access Extract, Collect Clean, Transform Decide, ActAnalyze, Visualize Consolidated data Distilled essence Applied knowledgeVarious data Apps Services Queries …. Visualization Reporting Notification Dataatrest DB / DW Text, mail Sensors Multimedia Social, web Click streams Datainmotion
  10. 10. 9 Copyright 2016 FUJITSU Big Data infrastructure master plan Data Sources Analytics Platform Access Extract, Collect Clean, Transform Decide, ActAnalyze, Visualize Consolidated data Distilled essence Applied knowledgeVarious data Apps Services Queries …. Visualization Reporting Notification Batch Processing Platform Dataatrest DB / DW Text, mail Sensors Multimedia Social, web Click streams Datainmotion
  11. 11. 10 Copyright 2016 FUJITSU Big Data infrastructure master plan Data Sources Analytics Platform Access Extract, Collect Clean, Transform Decide, ActAnalyze, Visualize Consolidated data Distilled essence Applied knowledgeVarious data Apps Services Queries …. Visualization Reporting Notification Distributed Parallel Processing Dataatrest DB / DW Text, mail Sensors Multimedia Social, web Click streams  Distribute data  Move code to data  Shared Nothing  Scale-out on demand  Data replication  Affordable servers Datainmotion
  12. 12. 11 Copyright 2016 FUJITSU Big Data infrastructure master plan Data Sources Analytics Platform Access Extract, Collect Clean, Transform Decide, ActAnalyze, Visualize Consolidated data Distilled essence Applied knowledgeVarious data Apps Services Queries …. Visualization Reporting Notification Batch Processing Platform Dataatrest DB / DW Text, mail Sensors Multimedia Social, web Click streams Datainmotion
  13. 13. 12 Copyright 2016 FUJITSU Big Data infrastructure master plan Data Sources Analytics Platform Access Extract, Collect Clean, Transform Decide, ActAnalyze, Visualize Consolidated data Distilled essence Applied knowledgeVarious data Apps Services Queries …. Visualization Reporting Notification Batch Processing Platform Dataatrest DB / DW Text, mail Sensors Multimedia Social, web Click streams Fast Response Platform  Result of pre-processing  Smaller than initial data volumes  All essential information  Base for analytics Datainmotion
  14. 14. 13 Copyright 2016 FUJITSU Big Data infrastructure master plan Data Sources Analytics Platform Access Extract, Collect Clean, Transform Decide, ActAnalyze, Visualize Consolidated data Distilled essence Applied knowledgeVarious data Apps Services Queries …. Visualization Reporting Notification Batch Processing Platform Dataatrest DB / DW Text, mail Sensors Multimedia Social, web Click streams  No random read / write  Limited access possibilities (D)FS Datainmotion
  15. 15. 14 Copyright 2016 FUJITSU Big Data infrastructure master plan Data Sources Analytics Platform Access Extract, Collect Clean, Transform Decide, ActAnalyze, Visualize Consolidated data Distilled essence Applied knowledgeVarious data Apps Services Queries …. Visualization Reporting Notification Batch Processing Platform Dataatrest DB / DW Text, mail Sensors Multimedia Social, web Click streams  Existing skills  Structured data, row store  Size affects response time  Columnar store for acceleration (D)FS SQL DB Datainmotion
  16. 16. 15 Copyright 2016 FUJITSU Big Data infrastructure master plan Data Sources Analytics Platform Access Extract, Collect Clean, Transform Decide, ActAnalyze, Visualize Consolidated data Distilled essence Applied knowledgeVarious data Apps Services Queries …. Visualization Reporting Notification Batch Processing Platform Dataatrest DB / DW Text, mail Sensors Multimedia Social, web Click streams  Flexible data model (schema-less)  Designed for being distributed  Designed for scale-out  Data replication (D)FS SQL DB NoSQL Datainmotion
  17. 17. 16 Copyright 2016 FUJITSU Big Data infrastructure master plan Data Sources Analytics Platform Access Extract, Collect Clean, Transform Decide, ActAnalyze, Visualize Consolidated data Distilled essence Applied knowledgeVarious data Apps Services Queries …. Visualization Reporting Notification Batch Processing Platform Dataatrest DB / DW Text, mail Sensors Multimedia Social, web Click streams (D)FS SQL DB NoSQL In-Memory  Data and operations in RAM  Real-time analytics  Disk storage for persistence  RAM mirroring Datainmotion
  18. 18. 17 Copyright 2016 FUJITSU Big Data infrastructure master plan Data Sources Analytics Platform Access Extract, Collect Clean, Transform Decide, ActAnalyze, Visualize Consolidated data Distilled essence Applied knowledgeVarious data Apps Services Queries …. Visualization Reporting Notification Batch Processing Platform Dataatrest DB / DW Text, mail Sensors Multimedia Social, web Click streams Event Processing Platform Datainmotion  Define rules (condition, action)  Data streams flow through rules  Latency, throughput  Distributed engines (NW edge)  Local decisions in real-time Fast Response Platform IoT
  19. 19. 18 Copyright 2016 FUJITSU Big Data infrastructure master plan Data Sources Analytics Platform Access Extract, Collect Clean, Transform Decide, ActAnalyze, Visualize Consolidated data Distilled essence Applied knowledgeVarious data Apps Services Queries …. Visualization Reporting Notification Batch Processing Platform Event Processing Platform DataatrestDatainmotion DB / DW Text, mail Sensors Multimedia Social, web Click streams Fast Response Platform
  20. 20. 19 Copyright 2016 FUJITSU How to Get to Your Infrastructure without Paralysis
  21. 21. 20 Copyright 2016 FUJITSU Building a data center infrastructure is complex Tasks to be completed  Select components from myriad of options  Procure, integrate  Sizing of CPU, RAM, storage, NW  Extensive testing (compatibility, bottlenecks)  Deep knowledge and skills to cope with technologies  Coordination among admins  Error-prone, time-consuming, risky, expensive  High maintenance effort Is there a fast track to data center infrastructures?
  22. 22. 21 Copyright 2016 FUJITSU FUJITSU Integrated System PRIMEFLEX Definition  Pre-configured, pre-integrated and pre-tested combination of data center components  Servers, storage, network connectivity, software Characteristics  Own and partner technologies  Optimally designed, based on best practices & experience  Traditional converged and hyper-converged  Ready-to-run and reference architectures  Long track record, many customer references  Addressing use cases of high relevance Reduce complexity, time, risk and cost.
  23. 23. 22 Copyright 2016 FUJITSU FUJITSU Integrated System PRIMEFLEX – Your fast track to data center infrastructures DESIGN - Select from myriad of SW and HW products INTEGRATE - Assemble involving multiple admin domains TEST - Work off complex test matrix DEPLOY Integrate into production environment Do-it-yourself approachTime to production DEPLOY Integrate into production environment Integrated Systems approach DESIGN - Based on best practices INTEGRATE - Pre-integrated single order delivery TEST End-to-end quality assurance Projectstart Solution Development in factory At customer's site
  24. 24. 23 Copyright 2016 FUJITSU PRIMEFLEX meets Big Data Data Sources Analytics Platform Access Extract, Collect Clean, Transform Decide, ActAnalyze, Visualize Consolidated data Distilled essence Applied knowledgeVarious data DataatrestDatainmotion DB / DW Text, mail Sensors Multimedia Social, web Click streams PRIMEFLEX for SAP HANA PRIMEFLEX for Hadoop PRIMEFLEX for Hadoop PRIMEFLEX for Hadoop ... ...
  25. 25. 24 Copyright 2016 FUJITSU PRIMEFLEX for Hadoop  Ready-to-run and reference architectures  Cluster of PRIMERGY RX / CX in various sizes  For storage- and processing-intensive tasks  Scale-out for coping with large data volumes  Automated integration of new components  Rapid modeling (collect, analyze, visualize)  Self-service (open by business users)  In-memory (RDD) for speed  Real-time stream processing  Configuration tool Analytical apps / templates (free) Datameer (Visual Analytics) Hadoop (HDFS, MapReduce, …) Linux OS PRIMERGY RX / CX Network switches The easy way to get in touch with Big Data.
  26. 26. 25 Copyright 2016 FUJITSU PRIMEFLEX for SAP HANA  Single and multi node  Various sizes  1st SAP validated multi-node  Single node available as VM  TDI support  Components SAP-certified  HA / DR options  SAP-certified components  Addressing all use cases  ERP, BW, B1, S/4HANA, BW/4HANA Your fast track to real-time insights. Internal storage or ETERNUS JX ETERNUS DX, NetApp FAS NW switches Single Node Multi-Node SAP HANA DB Linux VMware vSphere PRIMERGY RX / PRIMEQUEST certified for SAP HANA
  27. 27. 26 Copyright 2016 FUJITSU SAP HANA Vora™  What it is  In-memory query engine  Plugs into Apache Hadoop and Spark  Interactive analytics on data in Hadoop data lakes  Off-load data to Hadoop data lakes  How Fujitsu can help  PRIMEFLEX for SAP HANA  PRIMEFLEX for Hadoop optimized for SAP HANA Vora™ PRIMEFLEX for SAP HANA and PRIMEFLEX for Hadoop complement each other. PRIMEFLEX for SAP HANA PRIMEFLEX for Hadoop “Voracious” HANA user  SAP HANA Vora™ Data lakes Bridge between HANA and Hadoop
  28. 28. 27 Copyright 2016 FUJITSU PRIMEFLEX: Supported by flexible service options Consulting Design (configuration & sizing) Deployment Integration into production environment Additional services (e.g. migration) Lifecycle management Solution Support DC Services (Managed DC, Managed Hosting) Any guidance and support you need.
  29. 29. 28 Copyright 2016 FUJITSU What PRIMEFLEX customers have achieved 80% reduction in provisioning new services 5x faster time to market 5x more services deployed per year Cluster setup within 30 min instead of 3 days 99% reduction in setup time 72x faster to operation 70% reduction in deployment cost 30% reduction of infrastructure cost 35% less annual operation expenses 30% savings on overall IT budget 25% reduction in power and cooling 40% reduction in DC footprint 45% improvement in resource utilization 90% reduction in unplanned downtime 40% less time needed for system maintenance 60% of IT shifted from maintenance to innovation From 0 to VM within 10 min Figures may vary across PRIMEFLEX line-up and projects 2x more productivity and revenue
  30. 30. 29 Copyright 2016 FUJITSU HPDA (High Performance Data Analytics)
  31. 31. 30 Copyright 2016 FUJITSU HPDA (High Performance Data Analytics)  Typical characteristics  Structured data  Single source  Smaller volumes  Demand for high computing performance  Modeling and simulation: Objectives  Validate theories and designs  Build & test virtual prototypes  Increase prototyping coverage and quality  …  Use cases across industries  Crash simulation (automotive)  Testing new materials (construction and building)  Drug development against cancer (pharmaceutical)  Exploring cosmic origins and the universe Complex scientific, engineering, analytics tasks to solve.
  32. 32. 31 Copyright 2016 FUJITSU HPC – Foundation for HPDA  Working principle  User defines jobs using ISV software  Handover job to head node  Head node distributes jobs to compute nodes  Parallel execution of jobs on compute nodes  Compute nodes return results to head node  Head node returns overall result to user  Parameters affecting cluster size  Size of object (model)  Mesh size (accuracy of model)  Expected processing time  Middleware to bring cluster to life and make HPC accessible LAN Shared storage (Parallel File System) … Head node (management) ManagementNW(IP) High-speedNW(IB) Few to 1000s of compute nodes (job execution) Job request …
  33. 33. 32 Copyright 2016 FUJITSU Parallel MW Scientific Libraries Compilers, other tools Parallel FS Co-processor support (GPGPU and XEON Phi) OS (RHEL, CentOS) and drivers HPC Gateway Workload Manager Cluster deployment & management Servers (PRIMERGY RX, CX, BX) Network Switches (IB) Storage (ETERNUS DX)HPCClusterSuite(HCS) ... PRIMEFLEX for HPC: Addressing HPDA  Various flavors  Ready-to-run & reference architectures  With and without apps pre-installed  Intel Cluster Ready certified  HPC Gateway = simplification of HPC  Desktop-like look & feel  HPC accessible for non-IT users  Job submission from hrs to min  Multi-app support  Multi-cluster locations
  34. 34. 33 Copyright 2016 FUJITSU Analytics and Cloud
  35. 35. 34 Copyright 2016 FUJITSU Big Data and HPDA: Delivered from the Cloud?  Cloud is the answer to infrastructure on demand  Ever increasing data volumes (Big Data)  Varying performance needs (HPDA)  Infrastructure alternately used for Big Data and HPDA  Infrastructure not permanently needed (at full extent)  Public or private or … ? Cloud can be the foundation for Big Data and HPDA. Public Cloud Hybrid Cloud Private Cloud
  36. 36. 35 Copyright 2016 FUJITSU PRIMEFLEX for Red Hat OpenStack  Reference architecture  Private Cloud IaaS  Based on Red Hat Linux OpenStack  Optional upper layer cloud management SW (Catalog / Workload / Cloud Watch Manager)  Optional hyper-scale ETERNUS CD10000  Deliver (Big Data / HPC) infrastructure on demand  Take advantage from design of PRIMEFLEX for Hadoop / SAP HANA / HPC Deploy flexible, open, cost-effective private cloud platform in most reliable way. RHEL OpenStack Platform Hypervisor Apps Upper layer cloud management PRIMERGY CX / RX Network switches ETERNUS CD / DX
  37. 37. 36 Copyright 2016 FUJITSU Summary
  38. 38. 37 Copyright 2016 FUJITSU Summary  Data analytics is key prerequisite for digital transformation  Changing the way companies make decisions, do business, succeed or fail  Infrastructure matters  Various concepts for different use cases  PRIMEFLEX – Fast track to DC infrastructures  Big Data  HPDA  Cloud  …  End-to-end services  Sourcing options Free your analytics from paralysis? Have a word with FUJITSU. Experience across industries Choice Digitalization with confidence
  39. 39. 38 Copyright 2016 FUJITSU Let’s have a look at our PRIMEFLEX video
  40. 40. 39 Copyright 2016 FUJITSU

×