60 likes | 178 Views
What We Learned: Enterprise Hadoop Trends. June 19, 2014. Most common Infrastructure needs. A scale-out compute platform that leverages commodity components but also doubles as an economically feasible persistent, high-availability platform.
E N D
What We Learned:Enterprise Hadoop Trends June 19, 2014
Most common Infrastructure needs • A scale-out compute platform that leverages commodity components but also doubles as an economically feasible persistent, high-availability platform. • Provide a scalable infrastructure with minimum costs. • Structured and unstructured data sets from various sources (NoSQL, web feeds, traditional SQL systems). • Mostly use to replace current data warehouse systems. • Capture and archival of various data sets including real-time feeds. • Provides a huge cost savings vs. traditional storage systems. • Data analysis across various data sources via a centralized platform. • Consists of operational data, retail/ecommerce systems. This is the dominant use case.
Hadoop Deployment Information Below are the average current deployment trends: • Already deployed to production 32% • Planning in the next 12 months 31% • Evaluation/Prototype Stage 37% Below are the average reason for choosing a distribution: • Management costs 28% • Storage costs 23% • Support 18% • Ecosystem support/options/tools 13% • Vendor partnerships and support 12% • Other 6%
Hadoop Infrastructure Information Below are the average current cluster setups: • The average development Hadoop cluster is between 20 – 60 nodes • The average production Hadoop cluster is between 100 - 1000 nodes Below are the new upcoming enhancements coming to Hadoop: • Rolling upgrades • Better Management/Analytics/Monitoring tools • Core HDFS - ACLs, Memory Pinning • Ambari • Blueprints – Defines what packages you will install in your cluster. • Stack Definitions – Allows you to define which packages you will use in your distribution. • View – Allow you to add custom screens to Ambari for the various Hadoop applications.
Business Challenges Solved by Hadoop • Improved customer satisfaction • Reduce time to develop/market new product/solution • Reduce cost of doing business • Gain a competitive advantage • Gain better visibility to travel and expense spending via enhanced reporting tools. • Centralized data store