Latency Measurement in Software Defined Datacenter
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current technologies inadequately address network latency measurement in virtualized environments of software-defined datacenters, which is crucial for troubleshooting and optimization but has been insufficiently addressed, especially since most applications run in these environments and high latency impacts user experience and business operations.
Innovation Solution
A method for identifying packet latency in software-defined datacenters by storing time values associated with packet processing operations and transmissions across multiple host computers, using controllers to process these values and identify latencies, including middlebox service operations, forwarding operations, and virtual/physical network interface card operations, with optional sampling and bidirectional forwarding detection packets for latency measurement.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If latency measurement is performed in virtualized environments using traditional physical network measurement approaches, then measurement coverage is limited, but implementation complexity increases when adapting to virtual network layers
Solution Approach 1:
The patent segments the latency measurement process into distinct components: time value collection at host computers, time value transmission to controllers, and latency calculation at controllers. This segmentation allows each component to be optimized independently while maintaining overall measurement accuracy in virtualized environments.
Solution Approach 2:
The patent introduces controllers as intermediary components that collect time values from multiple host computers and perform the actual latency calculations. This intermediary layer simplifies the measurement process by centralizing the computation logic and enabling coordinated measurement across the virtualized network without requiring complex modifications to each host computer.
2Measurement precision
If all packet processing operations are monitored to identify latency segments, then measurement completeness improves, but processing overhead increases
Solution Approach 1:
The patent implements preliminary action by having host computers collect and store time values at specific packet processing operations before transmitting them to controllers. This pre-collection approach ensures that all necessary timing data is captured in advance, enabling complete latency segment identification without requiring real-time processing overhead during packet forwarding.
3Measurement precision
If latency measurement is implemented across all host computers and packet flows, then measurement comprehensiveness improves, but system resource consumption increases
Solution Approach 1:
The patent applies partial action by allowing administrators to configure whether to collect time values for all packet processing operations or only for specific operations of interest. This selective collection approach enables comprehensive measurement capability when needed while reducing resource consumption during normal operations by monitoring only critical latency segments.
Data Source
AI summary
Some embodiments provide a method of identifying packet latency in a software defined datacenter (SDDC) that includes a network, several host computers executing several machines, and a set of one or more controllers. At the set of controllers, the method o receives, from a set of host computers, (i) a first set of time values associated with multiple packet processing operations performed on packets sent by a set of machines executing on the set of host computers and (ii) a second set of time values associated with packet transmission between host computers through the SDDC network. The method processes the first and second sets of time values to identify a set of latencies experienced by multiple packets processed and transmitted in the SDDC.