A tribute to smartening your network management

Opinion paper by Paolo Lucente, author of the open-source PMACCT

PMACCT was born fifteen years ago out of my own operational needs when I was working for the Italian CNRS. I needed to collect and de-aggregate data from the equipment but not just any data, the right data set. So I’ve developed the software that could extract the data and store them in a database like MySQL for instance. Over time more and more plug-ins were added, including message brokers. On the input side of things, it became possible to collect BGP, IGP and to put in place streaming telemetry. To this day PMACCT is the only collector in the Streaming Telemetry space that is really multi-vendors.

Operators and entreprises around the world widely implement PMACCT for two main reasons. The first upside is the scalability of the software. Whether you’re one of the largest CDNs, processing tens of Terabit/s or at home with your router, you can use PMACCT and scale up as your data grow. You can also partition your network and run multiple data samples to issue different correlations.

In addition to scaling, PMACCT is a tool upon which you can develop several user cases such as Business Intelligence, insight in traffic patterns, peering analysis and decisions, forensics of network events, capacity planning and traffic engineering.

However I had noticed a gap on the market. On one side, companies with talented people are able to spend the time to develop their user cases in-house. On the other side, there are companies which need just as much these processes to better manage their networks but don’t have the time nor the workforce to achieve their goals. In the data collection world, products available are never tailor-made and there’s a learning curve regarding their APIs that require time and dedication.

That’s why I’m happy to see Pragma Innovation’s approach of providing a visualization software that, with consultancy, is also developed to meet the specific needs of the client networks. As an example, Pragma Innovation is able to work with SNMP that operators still use while progressively transitioning to streaming telemetry.

Going forward, companies will need more than visualization. They want smart tools such as alerting thresholds and algorithms identifying patterns providing the alerts in advance. Pragma Innovation is building this intelligent alerting component as well.

In a third phase, I expect that machine-learning will introduce ways to instrument the network and to provide auto-healing. On PMACCT side, we intend to follow the industry trends, for example tracking down Streaming Telemetry, contribute where possible to standardization, for example to BMP (BGP Monitoring Protocol), and adding support for new and relevant message brokers.