Ascend Use Case
Unified Batch & Streaming
Ascend orchestrates batch and streaming data with a single architecture that results in the continuous enrichment of datasets at the point of usage. This unified approach eliminates the need for separate infrastructures and tools, resulting in significant savings in cost and time.
Reduction in Code
Simply specify your data sources, transforms, and destinations. Ascend handles the rest.
Automation
for common problems with streaming architectures (i.e. automatically handles late data)
Cut Maintenance
by eliminating parallel architectures. Spend more time building, less time plumbing pipelines.
How it Works
- Connect directly to any stream (Kafka, Kinesis, etc.) and Ascend will automatically collect and persist the data
- Specify streaming data transformation logic the same as you would for batch with SQL, Python, Java, or Scala
- Ascend automatically processes and aggregates new data in time window optimized micro-batches
- Unify with your historical and batch data using time window optimized joins, and Ascend ensures your pipeline runs efficiently
- Write the resulting data directly to a data warehouse, blob store, and more
What to Expect
- Data Optimization: Automated best practices for storing and optimizing your data. For example, Ascend handles the small file problem with streaming data, automatically compacting and compressing arriving data into files to sustain performance.
- Native Timestamp Partitioning: Ascend’s DataAware Intelligence actively profiles all data in order to partition it by the time of the embedded business event (or by other fields in the record) rather than the time of arrival. This technology solves the late data arrival problem and avoids inefficient reprocessing further along the pipelines.
- Automated Backfill: You know how when you find a bug in your pipeline, and you have to deploy the fix across both batch and streaming code bases, backfill all old calculations, and recompute all datasets? Ascend’s DataAware Intelligence eliminates that entire process. As soon as it detects your code changes, it automatically traces the lineage of the existing data versions, narrowly locates only those partitions and records that are affected, and updates everything for you. Regardless of stream or historical data, the animated UI follows the processing in realtime and notifies you of progress along the way.
