Elevation-based Navigation System (EleNa)

Authors:

• Mudit Chaudhary (32607978)
• Pragyanta Dhal (33069605)
• Ayushe Gangal (33018381)

Introduction

EleNa is a web application providing elevation guided navigation i.e., it tries to minimize the elevation gain for generating directions between the origin and destination on the map. The web application gives the user functionality to choose the origin, destination, and tolerance (how much extra are they willing to walk for a lower elevation gain path). The web application then provides the optimized route between the two addresses.

Current constraints

Due to limited computational resources, we have the following constraints:

1. The origin and the destination addresses must be in the same city and state.
2. We limit the navigation to addresses in the USA.

How to build and start the web application

The web application follows a MVC microservices architecture. Each microservice runs on a separate docker container. We use docker compose for container orchestration and build scripts. We have the following services (docker_compose.yaml):

1. view: Web application interface for the user.
2. worker: Handles map downloads, processing, and finding the shortest routes. This can be scaled to more workers to handle more demand.
3. load balancer: Handles load balancing to multiple workers from view using Nginx.
4. lru_cacher: Runs a cron job every midnight to delete least recently used caches for maps and graphs.

To build and deploy the app, the developer needs to have docker installed. The developer also need to add their Google Maps Elevation API key in worker/config.json. More information about how to generate API key here

Step 1: Build service

To build the service, developer need to be in the root directory with docker-compose.yaml
Command: docker compose build

Step 2: Start the service

To start the service, developer needs to specify the scaling factor for the worker. Scaling requires change in load balancer configuration; the required changes are specified in later sections. By default, we scale to 2 workers.
Command: docker compose up --scale worker=2

Step 3: Scale worker process (Optional)

To scale the workers, load_balancer/nginx.conf needs to be modified. For example, to scale to 2 workers (default), the configuration is:

events { worker_connections 1024; }

http {

    # List of application servers
    upstream api_servers {
        server cs520_elena_map-worker-1:9999;
        server cs520_elena_map-worker-2:9999;
    }


    # Configuration for the server
    server {

        # Running port
        # listen [::]:5100;
        listen 5100;

        # Proxying the connections
        location /download_graph {
            proxy_pass         http://api_servers/download_graph;
        }
        location /get_shortest_path {
            proxy_pass         http://api_servers/get_shortest_path;
        }

    }
}

To scale the service to 5 workers, the configuration needs to be modified to:

events { worker_connections 1024; }

http {

    # List of application servers
    upstream api_servers {
        server cs520_elena_map-worker-1:9999;
        server cs520_elena_map-worker-2:9999;
        server cs520_elena_map-worker-3:9999;
        server cs520_elena_map-worker-4:9999;
        server cs520_elena_map-worker-5:9999;
    }


    # Configuration for the server
    server {

        # Running port
        # listen [::]:5100;
        listen 5100;

        # Proxying the connections
        location /download_graph {
            proxy_pass         http://api_servers/download_graph;
        }
        location /get_shortest_path {
            proxy_pass         http://api_servers/get_shortest_path;
        }

    }
}

After modifying the configuration, we need to run the build command again specified in Step 1. Then we deploy the service using a modified command specified in Step 2.
Command: docker compose up --scale worker=5
General command: docker compose up --scale worker=<scaling factor>

Step 4: Using the web application

The web application can be accessed on 127.0.0.1:5100. The screenshot below shows the expected interface.

Step 5: To stop the service

Command: docker compose down

How to run unit tests

To run the tests with ease, we provide a separate docker compose file (```docker-compose_test.yaml``). To run the tests, user needs to run:

1. docker compose -f docker_compose_test.yaml build
2. docker compose -f docker_compose_test.yaml up --scale worker=2

This will run two containers for testing the view (cs520_elena_map-view_test-1) and workers (cs520_elena_map-worker_test-1). The unit tests result can be viewed in the logs of these containers using Docker GUI or command line. The unit test results are shown below. Worker tests

View tests

Shut down the docker containers after tests are completed using: docker compose -f docker_compose_test.yaml down

Setup for API stress testing

We provide a separate docker compose (docker_compose_stress_test.yaml) for stress testing the worker internal API. It exposes the worker service on 127.0.0.1:5100 with two available end points /download_graph and /get_shortest_path. To perform setup for stress tests, user needs to run:

1. docker compose -f docker_compose_stress_test.yaml build
2. docker compose -f docker_compose_stress_test.yaml up --scale worker=2

After that the user can use a stress tester (e.g., Apache JMeter) for performing the testing.

We provide example cURL commands for making requests to the worker API:

• /download_graph

curl -X POST 127.0.0.1:5100/download_graph -H 'Content-Type: application/json' -d '{"city":"Amherst", "state":"MA","transport_mode":"walk"}'

• /get_shortest_path

curl -X POST 127.0.0.1:5100/get_shortest_path -H 'Content-Type: application/json' -d '{"city":"Amherst", "state":"MA","transport_mode":"walk", "from_address":"Brett Hall, Amherst, MA", "to_address":"497 East Pleasant Street, Amherst, MA", "tolerance":20.0}'

Shut down the docker containers after stress tests are completed using: docker compose -f docker_compose_stress_test.yaml down