Hierarchical In-Memory Binning for Low-Latency Telemetry Queries
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In-memory analytics systems face challenges in scaling memory capacity to accommodate large telemetry datasets and handling multiple queries with significant latency due to high volumes of data and query demands.
Innovation Solution
A hierarchical arrangement of bins in memory is used, each with a statistical metric based on data aggregates, allowing queries to be processed efficiently by identifying the appropriate hierarchical level for computations, maintaining a constant memory footprint through statistical metrics, and employing dynamic summarization and eviction schemes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If in-memory analytics systems store large telemetry datasets, then data availability and query accuracy are improved, but memory capacity requirements and system latency increase
Solution Approach 1:
The patent segments the in-memory data structure into multiple bins organized in a hierarchical arrangement with different aggregation levels. Each bin stores summarized representations of telemetry data at specific granularities, allowing the system to balance between storing detailed data for accuracy and using aggregated data to reduce memory consumption. This segmentation enables efficient memory utilization while maintaining query accuracy through selective access to appropriate aggregation levels.
2Quantity of substance
If in-memory analytics systems accommodate large telemetry datasets, then data coverage is improved, but query processing latency increases
Solution Approach 1:
The patent implements preliminary action by pre-computing and storing summarized representations of telemetry data in multiple bins at different hierarchical levels before queries are executed. When a query arrives, the system can immediately access pre-aggregated data at appropriate levels rather than computing aggregates on-demand, significantly reducing query processing latency while maintaining the ability to handle large datasets.
3Productivity
If in-memory analytics systems process multiple queries with high data volumes, then query throughput is improved, but system latency increases
Solution Approach 1:
The patent applies local quality by organizing bins in a hierarchical structure where different regions of the data structure serve different query types and granularities. The system can select appropriate bins at different hierarchical levels based on query requirements, allowing high-throughput processing of aggregate queries using coarsely-aggregated bins while maintaining accuracy for detailed queries through access to finer-grained bins, thereby reducing overall system latency.
Data Source
AI summary
In some examples, a system receives data values related to a computing environment, and produces, in a memory, an in-memory representation of the data values comprising a hierarchical arrangement of bins, where a bin of the bins includes a statistical metric based on an aggregate of a subrange of the data values, and the bins at different hierarchical levels of the hierarchical arrangement of bins represent different resolutions at which the data values are aggregated. The system receives, a query requesting a computation, the query comprising a query filter. The system identifies a hierarchical level of the different hierarchical levels of the hierarchical arrangement of bins based on the query filter, and performs the computation using bins at the identified hierarchical level.