Legacy data warehouses are no longer meeting today's need for fast, ongoing access to detailed analytics. Instead, organizations need a new type of data warehouse based on two-tier Lamba architecture accelerated by in-memory technology. This data warehouse of the future offers the performance they need while keeping costs low. The easiest way to transition to this data warehouse of the future is a four-step process. Enterprises often begin by offloading their ETL processes to Hadoop, resulting in significant cost savings. Once they have a Hadoop cluster up and running, the next logical steps are to expand their use of Hadoop by creating an active archive and then a data repository. From there, they can transition to two-tier Lamba architecture that gives them the ability to analyze both batch and streaming data. This tiered architecture offers the best of both worlds. The Hadoop-based batch tier provides the reliability organizations need for their financial reporting and can scale out over time. The speed tier, which leverages in-memory technology and tools like Spark, Storm, and Flink, maximizes performance and allows organizations to scale up while staying within budget. This whitepaper details the benefits of building this data warehouse of the future as well as offering tips for transitioning to Lamba architecture.
|