Dynamic Hierarchical Database Placement for Resilient Failover
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In clustered environments, such as datacenters or cloud computing, the optimal placement of pluggable databases across container databases is challenging due to imbalanced resource distribution, leading to operational and performance issues during failovers, as existing heuristic reconfiguration algorithms tend to be suboptimal, resulting in increased database latencies and decreased system throughput.
Innovation Solution
The implementation of a placement score system that calculates a numeric scalar to optimize resource allocation by forecasting the impact of failures and using linear or non-linear programming techniques to maximize resource utilization, allowing for proactive anticipation and mitigation of potential failures, thereby selecting the most resilient configuration.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If heuristic reconfiguration algorithms are used for database placement, then implementation simplicity is maintained, but resource utilization becomes suboptimal and system throughput decreases
Solution Approach 1:
The patent transforms the database placement problem from a combinatorial optimization challenge into a linear programming problem by changing the parameter representation to a placement score based on resource metrics. This allows the use of efficient linear programming solvers instead of complex heuristic algorithms, achieving both implementation simplicity and optimal resource utilization that maximizes system throughput
Solution Approach 2:
The patent replaces the mechanical heuristic search process with a mathematical linear programming system. By substituting the trial-and-error heuristic approach with a formal optimization framework using objective functions and constraints, the system achieves superior resource allocation and throughput while maintaining computational tractability
2Reliability
If multiple container databases host pluggable databases in clustered environments, then high availability is achieved, but optimal placement becomes a hard combinatoric problem
Solution Approach 1:
The patent changes the parameter representation from discrete placement decisions to continuous placement scores based on resource metrics. This transformation converts the hard combinatoric problem into a linear programming problem that can be solved efficiently, maintaining high availability requirements while reducing optimization complexity
Solution Approach 2:
The patent segments the complex placement problem into individual linear constraints and objective functions that can be processed separately by a linear programming solver. By breaking down the combinatorial complexity into manageable linear components, the system achieves optimal placement without overwhelming computational complexity
3Reliability
If failover occurs to maintain high availability, then service continuity is ensured, but resource distribution becomes imbalanced causing performance problems
Solution Approach 1:
The patent incorporates resource utilization metrics and performance data into the placement score calculation, creating a feedback mechanism that continuously adjusts placement decisions. This ensures that failover operations maintain service continuity while preventing resource distribution imbalances that would degrade performance, as the linear programming optimizer recalculates optimal placement based on current resource states
Data Source
AI summary
Herein are resource-constrained techniques that plan ahead for resiliently moving pluggable databases between container databases after a failure in a high-availability database cluster. In an embodiment that has a database cluster that hierarchically contains many pluggable databases in many container databases in many virtual machines, a computer identifies many alternative placements that respectively assign each pluggable database instance (PDB) to a respective container database management system (CDBMS). For each alternative placement, a respective placement score is calculated based on the PDBs and the CDBMSs. Based on the placement scores of the alternative placements, a particular placement is selected with a best placement score that indicates optimal resilience for accommodating adversity such as failover and overcrowding.