Hi,

Further to the earlier response, where we established that 500kbps would be a good baseline for the inter-node and client traffic.

Now we have to add the data stream to collect data from 250 devices as well.  Again, this really depends on which devices and what type of protocol they use.  Devices feature all kinds of interfaces, ranging from quite bandwidth efficient binary (usually proprietary) protocols, to standard SNMP, or HTTP / XML / JSON type of exchanges.

We'll have to make assumptions.  So let's assume SNMP.  Then it still depends on details such as whether or not the device supports bulk getting metrics, or if you have to fetch each metric individually.   And then of course also how many metrics there are, how fast you want to update those, etc.  Below you can find a quick example: 250 devices, getting 50 metrics each 5 seconds, 100 metrics each 60 seconds (the less critical data), and another 100 metrics each 3600 seconds.  That amounts to about 2.3 Mbps.  This would have to be added to the 0.5 Mbps that we already defined.

I'll ask to add the Excel to Dojo and to give you a link, just in case you want to play with it.  It's a basic model, but at least gives you something to go by.

Note: DataMiner also has a timer base control, which you can use to dynamically accelerate or decelerate the polling.  You could easily add a rule that for example slows down the polling, and ramps it up when a device is in an alarm state.  Not sure if that would be needed, because I guess that depends on how expensive that inter-site connection is for you.

I know this is not an exact response, but there are many variables here that we don't know, so this is the best we can do for now.