In today’s always-connected economy, businesses increasingly need to provide data-centric real-time services to customers, such as recommendations, targeted advertising and fraud detection. To make these experiences possible, companies have started transitioning from big data analytics, where data is processed in batches after collection, to fast data analytics, where data is processed in real-time to provide immediate insights to companies and their customers.
The SMACK stack–Apache Spark, Mesos, Akka, Cassandra, and Kafka–is establishing itself as a standard for these fast data architectures.
As the SMACK stack emerges as an industry standard, it is evolving in multiple dimensions:
Individual components: Even though the name SMACK is derived from specific components (which are themselves maturing and gaining functionality over time), in practice, data architects will swap individual components to of the stack to fit their specific needs. Consider for example, using Flink for stream processing, or ElasticSearch for storage.
New use cases: While often initially considered for fast data processing, we are seeing SMACK users take advantage of the characteristics of the stack such as platform elasticity for example for exciting new use cases.
In this talk, we will discuss how the properties of the SMACK stack can help you rethink your architecture, with an emphasis on the impact of current and future changes to the stack and its components.