Network Statistics Service

This repository contains a small python package which generates network statistics. It has built-in functionality to upload those statistics to a DynamoDB table but also provides an interface which can be implemented to publish this statistics to a service of your choice.

Installation

This package can be installed with pip install network-statistics-service.

Implementation Details

The package uses a command line executor to run the ping command for latency statistics and the speedtest comand (included in the package) for speed statistics.

Usage

The following is a basic example in which network statistics are returned as a dictionary.

from NetworkStatisticsService.statistics_service import StatisticsService

if  __name__ == "__main__":
    statistics_service = StatisticsService()
    stats = statistics_service.generate_statistics()

The stats variable will store a dictionary similar to the following: {'Loss': 0.0, 'Min': 24.244, 'Avg': 24.679, 'Max': 25.293, 'Stddev': 0.455, 'Download': 184.79, 'Upload': 280.34, 'Timestamp': 1591373792}. The keys are of string type and the values are of float type.

In case of an error, the returned dictionary will contain a timestamp and a string with the error message.

Disabling certain statistics

If your application does not need either the latency or the speed statistics, they can be individually disabled using the following properties on the StatisticsService object.

• speed_statistics_enabled - defaults to True;
• latency_statistics_enabled - defaults to True.

In the following example, the speed statistics are disabled.

from NetworkStatisticsService.statistics_service import StatisticsService
    
if  __name__ == "__main__":
    statistics_service = StatisticsService()
    statistics_service.speed_statistics_enabled = False
    stats = statistics_service.generate_statistics()

In this case, the stats dictionary will look like this: {'Loss': 0.0, 'Min': 20.366, 'Avg': 24.064, 'Max': 26.564, 'Stddev': 2.925, 'Timestamp': 1591374439}.

Configuring the ping count

By default, the service runs the ping command with a count of 20. This default value can be provided to the constructor: statistics_service = StatisticsService(default_ping_count=10)

In this case, all subsequent calls to the generate_statistics method will use a ping count of 10.

Another way of changing the ping count is on each call to generate_statistics: stats = statistics_service.generate_statistics(override_ping_count=5). This value overrides the value provided in the constructor.

Uploading the Network Statistics

The StatisticsService also provides means to upload the statistics dictionary and error logs. It comes with built-in functionality to upload to DynamoDB.

Configuring the StatisticsService for DynamoDB Upload

You can opt to upload either the statistics or the error logs, or both. This behaviour is controlled by the following properties of the StatisticsService object:

• stats_upload_enabled - defaults to False;
• logs_upload_enabled - defaults to False.

Setting your AWS Credentials

Credentials have to be provided via named profiles in credential files. Please check this documentation for information on how to achieve that.

DynamoDB Table Requirements

In order for the service to function, your DynamoDB table configuration must be able to support the generated data. Your DynamoDB table should have one configured property as hash key called Timestamp. If you use Terraform for provisioning infrastructure, you can use this file as an example.

Providing DynamoDB Information to the Statistics Service

The configuration is passed to the StatisticsService object via the DynamoDBConfig object. It provides the following properties:

• aws_profile - the AWS profile (from the credential files) to use. Defaults to default;
• aws_region - the AWS region in which your DynamoDB tables reside. Mandatory field.
• stats_table - the name of the DynamoDB table where statistics should be uploaded. Required only if stats_upload_enabled is set to True.
• logs_table - the name of the DynamoDB table where error logs should be uploaded. Required only if logs_upload_enabled is set to True.

The following example shows how to use the DynamoDB upload functionality.

from NetworkStatisticsService.statistics_service import StatisticsService
from NetworkStatisticsService.dynamo_db_config import DynamoDBConfig

if  __name__ == "__main__":
    statistics_service = StatisticsService()
    dynamo_config = DynamoDBConfig(aws_region="eu-central-1", 
								   stats_table="stats-table",
								   logs_table="logs-table")
    statistics_service.stats_upload_enabled = True
    statistics_service.logs_upload_enabled = True
    statistics_service.dynamo_db_config = dynamo_config
    statistics_service.generate_statistics()

In case the DynamoDB configuration is not valid, the generate_statistics method will raise an exception of type DynamoDbException.

Implementing your own Statistics Uploader

The service provides the possibility of implementing a custom class which can be provided to the StatisticsService object to be used in place of the built-in DynamoDB uploader.

Your class must implement the InformalStatisticUploaderInterface provided in the package. The object can be passed to the StatisticsService via the constructor. The following example demonstrates implementing a custom class for upload.

from NetworkStatisticsService.statistics_service import StatisticsService
from NetworkStatisticsService.informal_statistics_uploader import InformalStatisticUploaderInterface

class CustomStatisticsUploader(InformalStatisticUploaderInterface):
    def upload_statistics(self, stats_dict: dict):
	    print(stats_dict)
	    
	def upload_logs(self, logs_dict: dict):
		pass


if  __name__ == "__main__":
	custom_stats_uploader = CustomStatisticsUploader()
	statistics_service = StatisticsService(stats_uploader=custom_stats_uploader)
	statistics_service.stats_upload_enabled = True
	statistics_service.generate_statistics()

If your custom class does not implement the interface, the StatisticsService constructor will raise an exception of type ValueError.

Comments

If you have any questions regarding this Python package, do not hesitate to contact me.