SaaS Traffic Clustering for SD-WAN Path Probing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The growing number of software-as-a-service (SaaS) applications and their data centers poses a challenge for software-defined wide area networks (SD-WANs) as they require an increasing number of probes for application-aware routing, leading to scalability issues and inefficiencies in path probing.
Innovation Solution
The technique clusters traffic characteristics of SaaS applications into groups based on similar characteristics, allowing for a reduced number of probes to be sent collectively across these clusters, rather than individually for each application, and uses these probe results to inform routing decisions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If path probes are sent individually for each SaaS application, then application-aware routing accuracy is improved, but the number of probes and system overhead increase
Solution Approach 1:
The patent groups multiple SaaS applications into clusters based on their traffic characteristics and probes each cluster with a single representative probe. This merging approach reduces the total number of probes from one per application to one per cluster, directly addressing the contradiction between routing accuracy and probe quantity.
Solution Approach 2:
The patent creates representative traffic characteristics that universally represent multiple applications within a cluster. This single representative profile serves multiple applications simultaneously, allowing one probe to provide routing information for many applications, thus reducing probe overhead while maintaining routing decision quality.
2Adaptability or versatility
If the number of SaaS applications increases, then service coverage is improved, but the scalability of path probing deteriorates
Solution Approach 1:
By clustering applications with similar traffic characteristics, the system can scale to support more applications without proportionally increasing probe complexity. The clustering mechanism groups applications so that one probe serves multiple applications, making the system scalable.
Solution Approach 2:
The patent performs preliminary clustering of applications based on their traffic characteristics before probing. This pre-grouping action organizes applications into manageable clusters, allowing the system to scale by adding new applications to existing clusters or creating new clusters only when necessary, rather than requiring individual probe configurations for each application.
3Measurement precision
If probes are sent for every application, then routing decision accuracy is improved, but network overhead increases
Solution Approach 1:
The patent combines multiple application probes into single cluster probes, reducing network overhead while preserving routing decision accuracy. The representative traffic characteristics ensure that the reduced number of probes still provides sufficient information for accurate routing decisions across all applications in the cluster.
Solution Approach 2:
The patent creates representative traffic characteristic copies that embody the essential probing needs of multiple applications. Instead of sending actual probes for each application, the system uses these representative copies to infer routing information for the entire cluster, reducing network overhead while maintaining decision accuracy.
Data Source
AI summary
In one embodiment, a device clusters traffic characteristics of traffic associated with a plurality of online applications into one or more clusters. The device determines representative traffic characteristics for a particular cluster in the one or more clusters. The device generates, based on the representative traffic characteristics, a probing strategy for the plurality of online applications associated with the particular cluster. The device causes path probes to be sent along one or more network paths in accordance with the probing strategy.