Hadoop/BigData, IaaS, PaaS Behind the Hype, Real Use-Cases for Your Business Peter Ackermann Senior IT Consultant
Agenda Introduction Today s hype about cloud-services Infrastructure as a Service (IaaS) Plattform as a Service (PaaS) Hadoop/BigData Conclusion 3 BT Germany 3
BT Germany globally active, in Germany at home BT Germany is the German country organization of BT Global Services committed to serve customers in Germany: International network / service coverage in more than 190 countries For more than 15 years continuous presence in the German market Five BT branch offices Four own City Fiber networks Central Service Desk Stuttgart Hamburg Ratingen / Dusseldorf Eschborn / Frankfurt Munich About 7,000 customers (business customers, public institutions, network carriers, resellers, Internet service providers) all over the world Data center infrastructure linked around the globe Follow-the-Sun in technical operations and customer service Strong partnerships BT Germany 4
BT Germany The Modules of our Range of Solutions Integrated and innovative ICT solutions the approach of BT Germany:
Customers in Germany a selection In Germany, the customer base of BT includes about 1,000 companies and public institutions: BT Germany delivers matching solutions for many sectors: The customer base includes some of the biggest banks, insurers, chemical companies and technology groups. Numerous multinational groups More than two thirds of the DAX30 companies are customers of BT. BT Germany 6
Today s hype about cloud-services You can read about cloud in any media theses days What is missing, are facts about what cloud actually is BT Germany 7
Today s hype about cloud-services Cloud is nothing complete new it s an evolution of traditional IT evolution BT Germany 8
Infrastructure as a Service (IaaS) Cloud Clients Web browser, mobile app, thin client, terminal emulator,... Platform Application Infrastructure SaaS CRM, Email, virtual desktop, communication, games,... PaaS Execution runtime, database, web server, development tools,... IaaS Virtual machines. servers, storage, load balancers, network,... ( ) to provision processing, storage, networks, and other fundamental computing resources ( ) Source: NIST Definition of Cloud Computing BT Germany 9
Infrastructure as a Service (IaaS) Handling traditional IT-infrastructure consists of many, manual steps: call IT-department ( I need a server for XYZ ) detailed planning of hardware-requirements order hardware build hardware into rack configuration of network installation of OS and applications These steps can be error-prone and long running BT Germany 10
Infrastructure as a Service (IaaS) Insert heading here Downsides of using traditional IT Benefits from using IaaS Most servers are underutilized, typically there is an average of 10%-15% utilization Because of the low utilization, costs are very high in relation to utilization (rack space, current, cooling) Limited scalability Implementing high availability is complex Consolidation of workloads Automation reduces human-errors Central administration reduces costs Flexibility in scaling High availability can easily be implemented BT Germany 11
Infrastructure as a Service (IaaS) Caveats Not every workload is suitable to be virtualized Implementation of a virtual infrastructure has to be carefully planed, considering the needs of all workloads Proper capacity planning and continuous usagemonitoring is crucial BT Germany 12
Platform as a Service (PaaS) Platform Application Infrastructure Cloud Clients Web browser, mobile app, thin client, terminal emulator,... SaaS CRM, Email, virtual desktop, communication, games,... PaaS Execution runtime, database, web server, development tools,... IaaS Virtual machines. servers, storage, load balancers, network,... ( ) to deploy onto the cloud infrastructure consumer-created or acquired applications created using programming languages, libraries, services, and tools supported by the provider. ( ) Source: NIST Definition of Cloud Computing BT Germany 13
Platform as a Service (PaaS) In a perfect world, a developer: would have the exact same environment in development, staging and production would just use services like webservers, databases and storage could port his applications to different infrastructures without touching code had not to implement complex logic to make the application highly scalable (e.g. replicating data to new nodes) BT Germany 14
Platform as a Service (PaaS) PaaS can be a perfect world for a developer: Orchestrates underlying IaaS-layer Easy to use interface for developer Identical environment for development, staging and production Databases, storage, loadbalancers, are services, that can be consumed Application-monitoring built-in for automagic scaling and/or recovery of crashed applications Applications highly portable Complete stack for application livecycle (code, build, test, deploy, code, ) BT Germany 15
Platform as a Service (PaaS) Caveats Implementing a complete PaaS-stack can be complex Existing applications have to be adapted to a PaaSstack, but this can be minimal, as long as the applications were developed with some methodology in mind (see http://12factor.net) There is some learning involved how to use the hooks and services in a PaaS-environment, but usually developers are used to learn about new frameworks BT Germany 16
Hadoop/BigData ( ) Technologies and architectures, designed to economically extract value from very large volumes of a wide variety of data, by enabling highvelocity capture, discovery, and/or analysis. ( ) Source: IDC definition of big data BT Germany 17
Hadoop/BigData Traditional databasesystems vs. Hadoop/BigData systems: traditional database-systems are purpose built and optimized for record storage and retrieval with random r/w analytics need to scan through all data, random r/w is not needed scaling a traditional database-system to handle high-volume, high-velocity data can be very expensive to impossible Hadoop is built to handle high-volume, highvelocity data and runs on commodity-hardware BT Germany 18
Hadoop/BigData The amount of data that is generated on a daily basis will become more. The true challenge is, that most of this data is unstructured and the speed of change. Examples: A Boeing 747 produces up to 240TB of data on a single domestic flight Twitter had an average number of 500 million tweets sent per day in 2013 BT Germany 19
Hadoop/BigData Depending on your particular use-case, hadoop can help your business using Hadoop as cost-effective live-archive for your database outsource heavy, analytical jobs to Hadoop prequalifying data in Hadoop for your database aggregating different structured and unstructured data sources learning to ask new questions to your data deciding later, if particular portions of data are important BT Germany 20
Hadoop/BigData Caveats Hadoop is not a drop-in replacement for any database-system There are lots of tools beside the core-hadoopplatform. Finding the right ones for you will take time There are SQL-like query-tools, but they are limited compared to real SQL (e.g. no joins) BT Germany 21
Conclusion Lessons learned Using an IaaS-Infrastructure helps you to consolidate your workloads, achieve high availability and leverage your operational costs Using a PaaS-infrastructure helps you to speed up your development and make your applications portable Using Hadoop helps you to handle new requirements with data analytics and to support your legacy database-system Even Grandpa Simpson isn t afraid of the cloud BT Germany 22
Q&A Visit us at our booth at the hosting.fair BT Germany 23