LiveAnalytics Advanced Features
Explore how Amazon Timestream's LiveAnalytics uses scheduled queries and derived tables to improve time-series data reporting. Understand the design of rollups, pre-aggregation, and tiered architectures that enable fast, cost-effective dashboard reads and business reports. Gain insight into managing retention and late data for accurate, operational analytics within AWS.
With ingestion patterns, query tuning, retention life cycle, and cost modeling already covered in the previous lesson, the focus now shifts to a persistent operational challenge: dashboards and business reports that repeatedly scan massive volumes of raw time-series records are both slow and expensive. Amazon Timestream for LiveAnalytics addresses this with a native mechanism called scheduled queries, which precompute and store results so downstream consumers read lightweight, pre-aggregated tables instead of raw data. This lesson unpacks the mechanics behind scheduled queries, derived tables, rollups, pre-aggregation, and materialized reporting patterns. These features are why AWS recommends LiveAnalytics when you need low-latency, repeated aggregate reads over recent and historical time-series data. Understanding them is also essential preparation for comparing LiveAnalytics with Timestream for InfluxDB in the next lesson.
The cost of querying raw data
Every time a dashboard loads, it typically issues the same aggregation queries over the same time windows. Averages, counts, percentiles, and sums are computed on the fly by scanning memory-store or magnetic-store records. Each execution consumes compute resources and incurs per-byte query charges based on the volume of data scanned.
As data volume grows and more users or automated refreshes hit the same tables, costs multiply linearly while latency increases. A dashboard refreshing every 30 seconds across a fleet of 10,000 IoT devices can easily scan millions of rows per cycle, turning a simple monitoring page into one of the most expensive components in the architecture.
The desired state is the opposite: predictable subsecond reads at a fraction of the cost. The tipping point for choosing pre-aggregation is when dashboards need predictable, cheap, repeated access to aggregated results rather than ad hoc exploration of individual records.
Attention: Scheduled queries are not the same as ad hoc SQL statements, external ETL pipelines into Athena or Redshift, or simple retention policies. They are Timestream-native precomputation jobs that keep results inside the service for seamless querying with the same SQL interface.
A common exam distractor conflates these concepts. Recognizing that scheduled queries are a distinct, service-native feature is critical for both real-world design and certification scenarios. ...