Must have ISP Open Source & tools

Hey there

We are a growing ISP in Colombia and Latin America. I am interested in hearing from others regarding tools and software they recommend we must have such as LibreNMS, Rancid etc.

It’s greenfieldish now :wink: so feel free to recommend A-Z anything! :wink:

Hope this thread is useful others too!

Mehmet

* mehmet@akcin.net (Mehmet Akcin) [Mon 08 Jul 2019, 02:07 CEST]:

We are a growing ISP in Colombia and Latin America. I am interested in hearing from others regarding tools and software they recommend we must have such as LibreNMS, Rancid etc.

You should reach out to Euro-IX if you haven't already, every member IXP has documented what software they use in their switch database

  -- Niels.

My List:

Oxidized as a replacement for RANCID
Telegraf + InfluxDB = Tons of Grafana Dashboards
(Open Source Slack Alternative)
Ansible or Python Knowledge with Paramiko or netmiko for network automation.

BGP:

FRRouting - Mimics Cisco CLI
BIRD - Programming style config format.
Exabgp - Mostly used for API driven applications, monitoring with heartbeat scripts.
(many others)

DDoS detection and/or filtering:

Fastnetmon - Supports many methods for packet processing.
Ddosdetector (IPv4 Only) - Uses netmap for packet processing.

Top Talkers + Other Creativeness (like fib compressing, or route optimization):

pmacct - sflow/netflow combined with BGP, and a database backend

Servers:

Sensu or LibreNMS for Nagios type monitoring.

Diagnostics:

MTR - ...and knowing how to interpret it's output.

-Ryan

I don’t know if the areas have been evaluated or not. I would hyperconverge and virtualize as much as possible. I would attempt MPLS with a VRF gateway. Money will probably be an issue so hosting VoIP and Content services may be good. Are you using wireless, cable, satellite as the backhaul? If this is completely Greenfield, then evaluating a location, finding relay sites and etc should be done 1st.

Cyrus

Awesome list

We use https://cbackup.me/en/ over Rancid

This handles time-series data really, really well and also pairs quite well with the ELK stack (Elasticsearch + Logstash + Kibana) for event-oriented data. Kibana can talk to InfluxDB, and Grafana can talk to Elasticsearch to somewhat tie things together, but they each are of course focused on their own type of data.

In particular, Logstash is a great place to send your syslog streams from all your gear toward.

The Grafana guys have launched Loki
(https://github.com/grafana/loki#loki-like-prometheus-but-for-logs), which is
"Prometheus for logs".

Java as a dependency this day and age…

-Ryan

I would chime in with tools for network analysis and planning:

http://bgp.he.net/
http://isolario.it
http://radar.qrator.net

last one is something we work on as a community project.