MP4

Distributed ML model prediction service

Program to use a distributed group membership service with fault detection to batch process image queries and predict their outcomes using pre-trained image classification models

• Uses TCP to join network and UDP to communicate with other members
• Coordinator & Worker Architecture

Notes

This code builds on top of the previous code - the new files are inside

• src/membership
• src/conf
• src/cmd/coordinator
• src/cmd/process

Building for source

Generating executable from source code:

go build -o process

Running source file:

go run process.go

Running executable with input arguments:

./process <modelName> <batchSize>

Execution

Open Terminal and run these commands.

To run coordinator (typically on VM2):

cd cs-425-mp1/src/cmd/coordinator/

./coordinator

To run node process (on all VMs):

cd cs-425-mp1/src/cmd/process/

./process