To be honest I think we’re going a different way…
Basically use SW for monitoring, thresholding and alarms – but for data analysis, graphs etc, we skip SW.
SW doesn’t scale to the level for which we want to poll data – basically full interface mib of ~50k interfaces every minute, and loads of other stuff too (qos queues, netflow etc).
While SW can scale up on pollers, the bottleneck is the mssql database, so we’ll augment SW with respectively Grafana, Influxdb and this:
In a quick test, we’re running all 3 applications on the same VM as containers, and it seems to scale for our needs. If that changes, we can split the setup into multiple influxdb hosts to get more disk I/O. Docker allows
us a much easier install, as it will work on any Linux host…
So, while we could work on DS for SW, currently we don’t have any motivation for it…