ISSN: 2224-6274, RNPS: 0514, Vol. 21, No. 1, january - july, 2025.

Python tool for behavioral studies of

the IP Pools in the BRAS

Elio Ramón Avila Rodríguez, MSc.

Abstract

The deployment of postpaid services at interoperable hotel Wi-Fi

sites nationwide—allowing up to three devices per account—alongside

expanded coverage and higher occupancy, has increased IP address

utilization. Consequently, the Telecommunications Company of

Cuba S. A., ETECSA has experienced an increase in terms of requests

for IP address ranges at these sites, currently allocated via empirical

calculations based on estimated guest numbers and service usage. The

absence of tools providing accurate IP address usage statistics hinders

behavioral studies to determine optimal pool sizing. This may cause

service disruption and dissatisfaction for both customers experience

address exhaustion, while ETECSA faces inecient network resource

allocation. To address this, a Python-based utility was developed to

collect data for behavioral analysis. This tool was developed so as

to solve the this problem. This paper describes the work performed

and the results obtained with this tool, contributing to our Company

enhancing tourism-sector service delivery and optimizing network

resource eciency.

Keywords: Python, Netmiko, Network Automation, WiFi.

Received: 12/2023 | Approved: 02/2024 | Published: 08/2024

1 ETECSA. elio.avila@etecsa.cu

RESEARCH

pp. 24 - 33

ISSN: 2224-6274, RNPS: 0514, Vol. 21, No. 1, january - july, 2025.

Python tool for behavioral studies of the IP Pools in the BRAS

Elio Ramón Ávila Rodríguez, MSc.

Introduction

The service known as Wi-Fi Interoperability constitutes one

of the possible forms of Internet access provided by ETECSA, as

a service provider for hotels, campsites, restaurants, hospitals,

universities, public transport hubs, and many other places. This

service’s network architecture, which uses an IPoE solution with

Web authentication (Figure 1), customer trac is transported by

layer from the Wi-Fi network Access Points (AP) deployed at each

location, to the ME60 devices, operating as Broadband Remote

Access Servers (BRAS) in the IP/MPLS network. To accomplish this,

service VLANs are appropriately routed through the access switches

and transit equipment of the transport/aggregation networks, which

terminate at the corresponding BAS sub-interfaces on such BRAS.

User control is guaranteed in these latter devices through interaction

with the other system elements, such as: Session Control, Captive

Portal, AAA, and DNS. The BRAS also function as the DHCP servers

towards the customers, dynamically assigning IP addresses to their

Stations (STA) or terminal equipment. For this, it is necessary to

congure a private IP address pool, exclusive to each particular

site. ETECSA, as the service provider, manages and controls these

ranges, and their dimensioning is performed based on information

provided by the personnel where the service is installed, which

includes the estimated number of concurrent customers, calculated

from the number of installed APs and the expected occupancy of

the facilities, elements which can vary from time to time.

Figure 1. Summarized diagram of the Wi-Fi Interoperability service

pp. 24 - 33

ISSN: 2224-6274, RNPS: 0514, Vol. 21, No. 1, january - july, 2025.

In the designation of the above-mentioned Pool IPs may occur the

following cases:

1. Under-dimensioned pool: It is below actual needs. It may

happen that a number of customer devices request IPs and there are

no available addresses left to assign to them (the range congured

on the server is exhausted). This causes service impairment in those

customers who fail to get IP addresses.

2. Over-dimensioned pool: It exceeds the actual needs. It leads

to an inecient use of network resources (IPs) to the detriment of the

provider, which, in an extreme case, may be left without free ranges

for its operation.

3. Correctly dimensioned pool: It matches the actual needs.

There is no negative impact for either the customers or the provider.

As part of the Company’s prioritized attention to the tourism sector,

we are continuously working on improving the Wi-Fi service in hotels

and solving their demands. As a result, the Wi-Fi Interoperability

modality called Postpaid Wallet, which has been applied with good

results since 2022; started in Havana and has spread throughout the

country. Here are some new features of this service:

1.- Billing based on total trac processed and not on the credit

consumed from active accounts.

2.- Creation of customer accounts from the hotel itself, with access

to GESNAUTA platform.

3.- Possibility to connect up to 3 devices per account.

4.- Need for accounts to remain connected

as long as possible once authenticated. In order to achieve this,

the disconnection for low trac and inactivity was disabled in the

BRAS. In the particular case of hotels on the Gaviota chain, this last

option was set for disconnection at 15 minutes.

In this postpaid modality, customers’ accounts can remain active

until logout, disconnection that the AAA sends at 12 hours of being

authenticated. This, together with the expansion of WiFi service in

hotel facilities, with coverage in practically all areas and increased

levels of occupation, has brought with it greater use of IP addresses,

Python tool for behavioral studies of the IP Pools in the BRAS

Elio Ramón Ávila Rodríguez, MSc.

pp. 24 - 33

ISSN: 2224-6274, RNPS: 0514, Vol. 21, No. 1, january - july, 2025.

and thus the need to expand the Pool in those places where it is

required. In this context, there has been an increase in demands for the

expansion of the IP Pools by the hotels’ technical and administrative

sta, who make an estimate based on the number of guests according

to housing capacities, multiplied by the 3 possible devices to connect

per account, resulting in a high number of requested IP addresses

(mostly /22 and /21 ranges).

On the other hand, the provider’s NCE management system

provides indications of IP address exhaustion in the Pools, by

generating usage alarms above 80% of IP addresses (a congurable

value, it comes with a default value), but it does not provide in a

timeline the specic values it reaches, but only when they exceed the

congured threshold value. Therefore, in the cases under analysis,

the required information was obtained on a case-by-case basis, by

manually probing the BRAS where the specic service was set up.

This is an inconvenience for precise monitoring use of IP addresses,

which makes it dicult to carry out behavioral studies, that allow to

choose the optimal Pool and serve as evidence to show in response

to customer demand. This can lead to service impairments and

dissatisfaction, both for customers and our Company, a fact which

constituted the problem to be solved in our case and which motivated

the development of a scheduled utility on Python, with which it was

possible to collect the information needed to conduct the relevant

studies and analyses related to IP Pools.

Python is a trend in the world of data network automation.

(Damien, 2020), given by its simplicity, versatility, modularity,

database processing, file management and graphical interfaces

for viewing information, as well as by the multiple and powerful

libraries available for connection to equipment from various

manufacturers, as is the case with Netmiko (Byers, 2024), used

in this work. Furthermore, it has a large international community

of developers, which is active, participatory and collaborative.

These elements, as well as having previous experience of its use,

were crucial when choosing the programming language used. This

Python tool for behavioral studies of the IP Pools in the BRAS

Elio Ramón Ávila Rodríguez, MSc.

pp. 24 - 33

ISSN: 2224-6274, RNPS: 0514, Vol. 21, No. 1, january - july, 2025.

paper describes the work done and the results obtained with the

developed computer tool.

Materials and methods

This research was developed under an applied engineering and

research approach for development (R+D). The methodology was

structured in three sequential phases to address the core problem

of the lack of accurate data on the use of IP addresses in the Wi-Fi

Interoperability service. Phase I consisted of the analysis of possible

commands to be used on the BRAS ME60, which would allow

gathering the necessary information. Phase II focused on analyzing

the tools available for network automation that would allow remote

management of the BRAS via the SSH protocol. Phase III consisted of

the development and implementation of a utility for automated data

collection, processing and presentation. These phases are described

below.

Study of possible commands to obtain the required

information

As a rst task, a survey of technical information was carried out,

to know the available commands and to choose the one suitable for

dened purposes. Among them was chosen the command disp ip-pool

pool-usage pool-name (pool name), the output of which simply and

explicitly provides the required information, which is: Pool address

range, IPs in use (including those allocated, conicting and blocked)

and the percentage of use they represent.

Network automation with Netmiko Python library

As a second step, the possible tools for collecting the required

information were studied and the Netmiko library of Python was

chosen for this purpose. This library was developed by Kirk Byers

as an improved version of Paramiko and is designed to facilitate

automated interaction with multi-manufacturers’ network devices,

such as: Cisco, Juniper, Arista, and many others (Byers, 2024).

Netmiko uses several communication technologies, the most

important being the SSH (Secure Shell) protocol, which encrypts

the management information transmitted thus achieving safer

Python tool for behavioral studies of the IP Pools in the BRAS

Elio Ramón Ávila Rodríguez, MSc.

pp. 24 - 33

ISSN: 2224-6274, RNPS: 0514, Vol. 21, No. 1, january - july, 2025.

communication. SSH is available on most telecommunication

devices for their remote access, this being the most used method

when applying automation to the field of telecommunications

networks (Damien, 2020). This library helps network operators to

send commands to equipment such as switches, routers, firewalls,

among others, in a scheduled and repeatable way, and offers a simple

and consistent way to automate network device management tasks

(Figure 2). This minimizes the manual intervention by operators

and reduces the margin of human error. This fact, combined with

Python’s multi-area capabilities, were the reasons for deciding on

this tool when choosing an effective network automation method.

Figure 2. Network automation using Netmiko

Python tool for behavioral studies of the IP Pools in the BRAS

Elio Ramón Ávila Rodríguez, MSc.

Structure of the Python-based program

Subsequently, a program on Python was developed, consisting

of two modules (Figure 3); the rst one (left) shows a Graphical User

Interface (GUI), which allows to enter the necessary parameters for

its execution. In its development, it was used primarily in the Tkinter

library. The second one (right) has in turn two parts: one for the periodic

connection to the equipment via SSH (using the Netmiko and Schedule

libraries), where the chosen command, described above, is sent to the

BRAS. This allows to query one or several selected IP Pools and properly

process the information it returns. The other part, after nishing the

pp. 24 - 33

ISSN: 2224-6274, RNPS: 0514, Vol. 21, No. 1, january - july, 2025.

number of dened readings, stores the data obtained in les with .csv,

.txt and .pdf extensions, in the latter case shows them graphically. The

Pandas and Matplotlib libraries were used for this part.

Program execution steps

The program execution consists of four steps (Figure 4):

1. Choosing the desired BRAS from a drop-down list (Figure 4a).

2. Choosing the IP Pool(s) to be analyzed from a drop-down list (it

automatically obtains them by connecting to the selected BRAS)

(Figure 4b).

3. Entering the start time to take readings and the number of

readings to be performed (Figure 4c).

4. Clicking on the ‘Execute Script’ button (Figure 4c).

Once the above is done, module 2 starts running at the time entered;

it takes the number of desired readings and processes the obtained

information.

a) b) c)

Figure 3. Structure of the Python-based program

Figure 4. Program execution steps

Python tool for behavioral studies of the IP Pools in the BRAS

Elio Ramón Ávila Rodríguez, MSc.

pp. 24 - 33

ISSN: 2224-6274, RNPS: 0514, Vol. 21, No. 1, january - july, 2025.

Results and discussion

Once the execution cycle of the program as shown in the previous

section is nished, the information obtained is stored in Excel and

text les, with .csv and .txt extensions (an example of this is shown in

Figures 5 a and b, respectively), as well as graphs with a .pdf extension.

a) b)

The case in Figure 6 is one of those that was monitored based on

the alarms generated in the NCE management system. According to

the information obtained with the Python utility, it was veried that IP

address exhaustion was occurring (in the red circles, moments when

it reaches 100% use) and thereby aecting customers.

Figure 5. Example of Excel and text information files

Figure 6. Under-dimensioned IP Pool Behaivor

Python tool for behavioral studies of the IP Pools in the BRAS

Elio Ramón Ávila Rodríguez, MSc.

pp. 24 - 33

ISSN: 2224-6274, RNPS: 0514, Vol. 21, No. 1, january - july, 2025.

Figure 7 shows a case where the IP pool was increased to /21, in

respond to a hotel demand. It can be seen in the obtained graphs that

does not exceed 20% of the use of IP addresses, being an example of

over-dimensioning, with implications for our Company.

Python tool for behavioral studies of the IP Pools in the BRAS

Elio Ramón Ávila Rodríguez, MSc.

Figure 7. Over-dimensioned IP Pool Behaivor

Figure 8. Correctly-dimensioned IP Pool Behaivor

Figure 8 shows a correctly sized case. This particular one requested

to expand its IP address range, which was not necessary to do at that

time, based on the statistics obtained and shared with the hotel sta.

It should be mentioned that in all the examples shown, there was

a high level of occupancy in the hotels, which in turn suggests high

percentages of IP address use.

pp. 24 - 33

ISSN: 2224-6274, RNPS: 0514, Vol. 21, No. 1, january - july, 2025.

Conclusions

The conclusions of this research reect the positive impact of

implementation of a computer tool developed in Python to analyze and

manage the IP Pools within ETECSA’s network. First of all, it is shown

that this tool is an eective solution to overcome the lack of accurate

information, allowing for automated collection and processing of

behavior of IP Pools, especially in environments where BRAs act as DHCP

servers. This facilitated early detection of issues, such as exhaustion

of IP addresses in hotels, and allowed timely corrective actions to

be taken, such as expanding the address range, avoiding service

interruptions. In addition, it was found that a correct dimensioning

of the Pools avoids unnecessary costs in the acquisition of new

IP addresses, contributing to a more efficient use of resources.

The statistics generated not only improve service quality Wi-Fi

for users, but also provide valuable evidence for technical and

administrative staff, increasing customer satisfaction. Finally,

although the tool was initially designed for the hotel sector,

its application potential is extended to other network IP Pool

services, demonstrating its flexibility and usefulness in different

ETECSA’s network resource management contexts.

References

Garros, Damien (2020). NetDevOps Survey. https://dgarros.github.io/netdevops-

survey/reports/2020

Byers, K. (2024). Biblioteca Netmiko. https://github.com/ktbyers/netmiko

ME60 V800R023C00SPC500 Conguration Guide. Información técnica del

proveedor Huawei (2023). https://support.huawei.com/enterprise/en/doc/

EDOC1100335692/d0a3124b/conguration-management-conguration

pp. 24 - 33

Python tool for behavioral studies of the IP Pools in the BRAS

Elio Ramón Ávila Rodríguez, MSc.