Tiny Social Network System (TinySNS)

Overview

The TinySNS project is an assignment from CSCE 662: Distributed Processing System, focused on building a scalable, fault-tolerant social networking service. This project extends the SNS service built in MP2.1 by incorporating high availability and fault tolerance features.

Key Features

• Fault Tolerance and High Availability: The system is designed to handle failures transparently, ensuring continued service availability.
• Master-Slave Architecture: Each cluster has a Master-Slave pair for redundancy and fault tolerance.
• Follower Synchronization: Dedicated processes synchronize follower information across clusters.
• Centralized Coordination: A coordinator manages load balancing and server synchronization.
• Resilience Testing: The system includes specific test cases to check its resilience against failures.

System Architecture

The architecture of TinySNS consists of several key components:

• Coordinator: Manages load balancing and server process synchronization.
• Servers: Handle core SNS functionalities like FOLLOW, UNFOLLOW, TIMELINE, and LIST.
• Clients: Users interact with the SNS through these client processes.
• Follower Synchronizers: Ensure consistent view of user data across clusters.

Limitations

Currently, the system has certain limitations:

• UNFOLLOW command works only within the same cluster.
• TIMELINE mode requires manual updates for cross-cluster user posts. An input of UPDATE in timeline mode refreshes the timeline with all available posts from clients followed by the user.

Testing

The document provides detailed test cases to validate the functionalities and resilience of the system. These include sanity checks within clusters, cross-cluster functionality checks, and system resilience tests.

Getting Started

Instructions to run the coordinator, servers, synchronizers, and clients are detailed in the design document. Please refer to the test cases section for specific commands and procedures.

---

Usage

You might need to change the location of gRPC in the Makefile. Then running a simple make builds all the necessary executables and object files.

make

Then to run the coordinator, servers, and synchronizer processes, execute the following commands.

# Run the coordinator
./coordinator -p 9090

# Run the servers
./tsd -h localhost -p 3010 -s 1 -k 9090 # cluster 1
./tsd -h localhost -p 3011 -s 4 -k 9090

./tsd -h localhost -p 3012 -s 2 -k 9090 # cluster 2
./tsd -h localhost -p 3013 -s 5 -k 9090

./tsd -h localhost -p 3014 -s 3 -k 9090 # cluster 3
./tsd -h localhost -p 3015 -s 6 -k 9090

# Run the synchronizers
./synchronizer -n 2 -p 3029
./synchronizer -n 3 -p 3030
./synchronizer -n 1 -p 3031

# Run the clients
./tsc -u 1 -k 9090
./tsc -u 2 -k 9090
./tsc -u 3 -k 9090
./tsc -u 5 -k 9090