Solution for Monitoring Physical and Virtual Telecom Networks
Solution for Monitoring Physical and Virtual Telecom Networks
Information
Region:
Worldwide
Industry:
Telecom
Type:
Web
Engagement model:
Dedicated team
Duration:
6 months and ongoing
Staff:
6 software developers and a QA automation specialist
Technologies used
Java
Spring
JMS
hibernate
PHP
React
MFC
C/C++
MySQL
Ajax
Motif
Perl
PostgreSQL
Apache

Client

A large telecom software and hardware provider used Elinext to build an application for monitoring physical and virtual network components in real-time.

Challenge

Over the past few decades, the digital economy has been booming. The demand for online services has been growing, and so has the pressure on telecommunication networks. Providers responded by adding new, advanced technology to their systems, but that came at a cost.

At some point, telecom companies had to operate hybrid infrastructures where diverse, advanced software complemented hardware (that was sometimes very old-school). Monitoring such infrastructures was increasingly difficult and cost companies a lot of money.

Our client, a multinational telecom software and hardware provider, approached that challenge with a product upgrade idea. It was a gateway addon that would allow telecom companies to monitor software-defined networking (SDN) and network functions virtualization (NFV).

In other words, a solution that puts bare metal and digital components on one screen.

 

“Monitoring hybrid networks was increasingly difficult and cost telecom companies a lot of money.”

Elinext had been developing and supporting the product the client wanted to upgrade for about a decade. The product had changed hands a few times as a result of corporate acquisitions, but every new owner had chosen to continue relying on us. So, naturally,  we were entrusted with the upgrade.

Process

We approached upgrading the product the same way we had been developing it before: agile development split into quarterly iterations for each feature.

At different stages of the project, our team composition changed, reaching 20 people whenever challenges required. Those challenges included discovering new technology, client-specific implementation and testing and visualizing data in a new way.

Discovering New Technology

Hybrid network components are interconnected at the host level. And that’s different from the modern default, where system units talk via Docker REST API. Therefore, we had to study how host-based connections worked, how to get data from them, and how to visualize that data.

That inquiry led us to learn about VMWare’s NSX-T. This platform has been used by many telecom providers to build software-defined networks (SDN), which the addon was designated to monitor. So, we had to get a sense of how we could pair NSX-T with the existing system and its data processing model.

It took us going over mounds of documentation to understand the essential technology. But that was a crucial first step that laid the groundwork for customizing the product for each end client in the future. Without that groundwork, we would be spending much more time studying telecommunication providers’ networks today.

Implementation and Testing

The upgrade required a few changes made to the core of the original technology stack. The major change was migrating from Oracle to Vertica, and that’s pretty much where one-time upgrades end.

The product implementation isn’t a step that you do once. Every telecommunication company is different when it comes to its network specifics and requirements. And that means we need to use the groundwork we’ve laid with the product to build it anew for every end client. That process usually takes 1.5 months.

Another thing that we’ve developed to use in future customizations is automated tests. We modify them and run them in a network monitoring simulator to make sure all connections within the network perform as expected.

Visualizing Data in a New Way

The client produces a variety of devices. And network switches are the most important ones, as they ensure uninterrupted experiences for each peer in the network.

Every such switch has a chip at its heart, where its functionality is encoded. And digging up that functionality code for monitoring the switches wasn’t an easy task.

First of all, we had to pick apart the code that comes with new types of switches. Those switches sent data via REST API which the original system could not process. So, we worked closely with the client’s admins and engineers to find a way to convert that data and make it usable.

Visualizing the data in a new way required updating the frontend. And the original frontend ran on an outdated custom framework which was made as a single web page. Although we couldn’t afford to rewrite the framework, we still could develop new React components and integrate them with the old architecture.

 

“We built a new frontend on top of what had proven itself indispensable time and again.”

Now, this may sound like an easy ride, but that’s far from what transpired. In fact, adding contemporary to the outdated system challenged our expertise like nothing before. But it was worth it: we essentially built a new frontend on top of what had been there for a long time and already proved itself indispensable.

Integration and Customization

Every time an end client wants to use the product with their system, we start by studying the problem and specifying the task. As part of that inquiry, we analyze the client’s network. And that means understanding how specific components work, how to integrate them, and how to get data from them in a digestible format.

Then, as previously mentioned, we build a customized product atop the basis we laid out earlier.

 

Product

The hardware and software our client supplies form the network backbone for many telecom providers. If anything gets broken, the end user will suffer interruptions in streaming, gaming, or more critical processes. And the product we helped build enables providers to spot issues in advance, across physical and digital dimensions.

 

Virtual Switches

Most issues arise at the physical, or the underlay level — on chips that are built-in boxes acting as physical switches. And those issues impact the overlay level — as represented by virtual switches running in clouds and receiving data from virtual machines and containers.

Usually, network admins only notice an error at the overlay level and cannot track down its underlay source. But the product we’ve built changes that.

We established a data exchange between physical and virtual components where admins can localize issues. Once the source of a problem has been discovered, the provider can send a technician to fix the physical switch causing an interruption.

 

“Usually, network admins cannot track down physical issues causing interruptions. The product we’ve built changes that.”

It’s worth mentioning that VMWare doesn't produce or support hardware. Nevertheless, we managed to visualize software-defined networks built using VMWare NSX-T and hardware indicators within one interface.

Web Interface

We established a meaningful connection between physical and virtual components. To visualize that connection, we developed a web interface.

The web interface is the most important part of the product. It includes hundreds of metrics network admins can track, alongside charts and advanced analytics. Essentially, we built a holistic topology of virtual components, displaying them within the interface alongside the physical ones.

In addition to plain switches, admins can use the application to monitor computers and routers that form a network from a user-friendly console. If a network component shuts down due to an error (e.g. when a switch CPU dies), system engineers will receive a notification immediately.

Network Monitoring Simulator

Customization is at the heart of the product. And it takes a lot of testing to ensure that the addon works as expected with an end client’s network. One way to ensure that is by pestering the end client for feedback, which can be bothersome for both parties. But we found a better solution: a network monitoring simulator.

We send each end client a utility program that they install in their system. And that’s all the telecom provider needs to do. The program creates a copy of their virtual network, and we can use that copy for autonomous testing.

 

"We create a copy of an end client’s virtual network and use it for autonomous testing.”

Once we’ve customized the product for the end client, we roll out the emulated environment. It allows us to run automated integration tests to ensure that connections between network components work properly. And the end client can go about their business as we do so.

Results

Our team helped pioneer a one-of-a-kind application pairing purely virtual and purely physical network switches in the same interface. And telecom providers benefit from it by securing uninterrupted online experiences for their customers whose demand is ever-growing.

“Our team helped pioneer a one-of-a-kind application pairing virtual and physical network switches in the same interface.”

1-96
2-85
3-69
4-52
5-45
6-26
7-21
8-17
9-8
Do you want the same project?
Got A Project Idea? Lets Discuss It With Us
Contact Us