Server Classification via Performance Signatures

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Traditional server classification methods are slow and inefficient in dynamic cloud environments, failing to keep up with rapid changes in server configurations and reconfigurations, and often require manual tagging or installation of specialized tools, which is impractical in shared infrastructure settings.

Innovation Solution

A computer-implemented method that collects server performance data over a time interval, generates a covariance matrix signature for each server, and compares it to known server types to classify servers automatically, using pre-defined rules and machine learning techniques to maintain a catalog of server signatures across multiple tenants.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If traditional server classification methods are used, then manual tagging or specialized tools can identify server types, but the classification process is slow and cannot keep up with rapid changes in cloud server configurations

Engineering Contradiction:

Improveserver classification speedVSAvoidtime to detect server type changes

Core Design Contradiction:

ProductivityVSLoss of time

Solution Approach 1:

The system enables servers to self-classify by automatically generating performance signatures from their own operational data and comparing them against a catalog, eliminating the need for manual tagging or external specialized tools. This self-service mechanism allows continuous automatic reclassification as servers are added, removed, or reconfigured in the cloud environment.

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The system pre-generates a catalog of server performance signatures for known server types before deployment. When new servers are introduced or existing servers are reconfigured, the classification process quickly matches their performance signatures against this pre-existing catalog, enabling rapid identification without time-consuming analysis from scratch.

Inventive Principle:

Principle #10Preliminary action

2Measurement precision

If specialized monitoring tools are installed on servers to enable classification, then server types can be identified, but this is impractical in shared infrastructure settings where access is restricted

Engineering Contradiction:

Improveserver type identification accuracyVSAvoidease of deployment in shared infrastructure

Core Design Contradiction:

Measurement precisionVSEase of manufacture

Solution Approach 1:

The system introduces a centralized signature catalog as an intermediary between the monitoring system and servers. Instead of installing specialized tools on servers, the system collects performance data through standard interfaces, generates signatures, and matches them against the external catalog. This intermediary approach enables accurate classification without requiring privileged access or specialized software on the servers themselves.

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The system replaces the mechanical approach of installing physical monitoring agents or specialized software on servers with a data-driven approach using performance metrics and statistical signatures. By substituting direct server access with analysis of performance data through standard interfaces, the system achieves classification capability without the deployment complexity of traditional methods.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

3Measurement precision

If comprehensive server performance data is collected to improve classification accuracy, then more metrics are available for analysis, but the complexity of data processing increases

Engineering Contradiction:

Improveclassification accuracyVSAvoidcomplexity of signature generation and comparison

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The system transforms raw performance data into standardized statistical parameters (mean, standard deviation, skewness, kurtosis) that capture the essential characteristics of server behavior. This parameter transformation reduces the complexity of processing comprehensive performance data while maintaining or improving classification accuracy, as the statistical parameters provide a compact and informative representation of server performance patterns.

Inventive Principle:

Principle #35Parameter changes

Data Source

PatentEP3503473B1Server classification in networked environments

Publication Date: 2021.01.06 BMC SOFTWARE INC

Figure 1A
Figure 1B
Figure 2

AI summary

A computer system (100) for classifying one or more servers by server type in a networked computing system (110) to institute server-type based monitoring and or maintenance of the networked computing system. The computer system (100) includes a processor, a memory, a data receiver, a server signature generator, and a server-type tagging service. The data receiver collects (103) server performance data for a first server over a time interval. The server signature generator (104) determines a signature of the first server based on the collected server performance data. The server-type tagging service (105) compares the signature of the first server to a signature of a second server of known server type, determines a similarity of the signature of the first server to the signature of the second server, and, based on the similarity, classifies the first server as being of the same server type as the second server.