Graph Workload Placement Using Telemetry-Aware Infrastructure Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing technologies struggle to efficiently place graph-based workloads on graph-based infrastructure topologies while satisfying service level agreements (SLAs), particularly due to the challenge of allocating workload tasks onto infrastructure nodes without adequate consideration of real-time infrastructure parameters and node health.
Innovation Solution
A method involving graph embeddings to represent workload and infrastructure topologies, using a historical database to find similar past executions, filtering candidates based on real-time telemetry, and assigning tasks with a 'northwest corner' algorithm for capacity planning, ensuring SLA compliance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If graph-based workloads are placed on graph-based infrastructure topologies using traditional allocation methods, then the placement process is simple, but SLA compliance deteriorates due to insufficient consideration of real-time infrastructure parameters and node health
Solution Approach 1:
The patent segments the placement process into distinct phases: (1) retrieving past workload executions and their infrastructure graphs from a database, (2) filtering candidate infrastructure subgraphs based on real-time telemetry data, (3) ranking candidates using a lie-within relationship, and (4) assigning tasks using a northwest corner algorithm. This segmentation allows complex SLA compliance to be achieved through systematic, manageable steps rather than a monolithic approach.
Solution Approach 2:
The patent performs preliminary actions by pre-storing past workload executions and their associated infrastructure graphs in a database before actual placement occurs. This historical data is retrieved and filtered in advance based on real-time telemetry, allowing the system to make informed placement decisions without requiring complex real-time computation during the actual workload placement.
2Measurement precision
If workload tasks are allocated onto infrastructure nodes without considering real-time parameters, then the allocation process is fast, but accuracy and SLA compliance deteriorate
Solution Approach 1:
The system performs preliminary retrieval of historical workload executions and their infrastructure graphs before actual placement. By pre-processing and storing this data, the system can quickly filter and rank candidates during placement without performing complex computations in real-time, thus maintaining both accuracy and speed.
Solution Approach 2:
The patent uses copies of historical infrastructure graphs and workload executions to inform current placement decisions. By retrieving and filtering past execution patterns, the system creates virtual models of potential placements that can be evaluated quickly against current SLA requirements without physically attempting every possible allocation.
3Reliability
If the workload placement system considers real-time infrastructure parameters and node health, then SLA compliance improves, but the complexity of assessing node states and network links increases
Solution Approach 1:
The patent introduces an intermediary ranking mechanism that uses a lie-within relationship to evaluate candidate infrastructure subgraphs. This intermediary function processes real-time telemetry data and historical patterns to produce a ranked list of candidates, simplifying the complex task of assessing multiple node states and network links into a manageable ranking process.
Solution Approach 2:
The system incorporates feedback from real-time telemetry data about infrastructure node health and network conditions. By continuously monitoring and using this feedback to filter and rank candidate subgraphs, the system adapts its placement decisions to current conditions without requiring complex manual assessment of each node state.
Data Source
AI summary
One example method includes maintaining a database of past executions of graph-based workloads and respective infrastructure graphs where the graph-based workloads were executed, obtaining a graph embedding of a new graph-based workload and searching for similar graph-based workloads in the database, retrieving each similar graph-based workload from the database, where each of the similar graph-based workloads is associated with a respective infrastructure subgraph corresponding to where that similar graph-based workload was executed, searching a large graph-based infrastructure topology for infrastructure subgraphs that are similar to the infrastructure subgraphs retrieved from the database, and retrieving a group of infrastructure subgraphs that collectively define a set of candidate nodes to run the new graph-based workload, filtering and ranking the set of candidate nodes according to a respective likelihood that the candidate nodes will satisfy requirements of the new graph-based workload, and assigning the new graph-based workload to one of the candidate nodes.