Power Strip Bandwidth limit?

[DCIM_Support]

The AP7930 Power Strip (PDU) sends roughly 880Bits/s from the PDU to DCE. I know that may not sound like a lot but if you have 100, 1000 or more power strips then the bandwidth adds up.

Is there any way to reduce the size of the data? Can I adjust to polling frequency and  not loose any data? Currently it is set to the default frequency.

Can I choose which data to send? ie. maybe I don't need the status of every outlet every second.

Also, what is the PDU sending? Just trend points or something else?

Regards,

J

(CID:93194071)

[DCIM_Support]

by default, DCE polls every device every 5 mins or when an alert trap is sent to DCE.  The data points DCE polls is determined by the snmp oids in the device definition file (ddf) and which ones the device responds too. 

You can adjust the polling frequency to something higher than 5 mins by adjusting the device polling settings.  It just depends on how often your data changes and how many data points you're looking to capture.  You cannot edit the ddf to adjust which points it polls or does not poll.

The pdu is sending all the points that show up when you right click the device and choose view device sensors.  When it is in an alert state, it is also sending alarm data.

(CID:93194076)

[DCIM_Support]

Firstly i would like to state that i don't know your particular situation so please don't shoot me down 😉 

But the network which polls these devices is a separated VLAN (or it should be). Based on a 100Mbit network, which is widely available these days, you can manage roughly 100.000 devices before you would get any bandwidth issues. And that is based on the fact that all the devices would be polled at precisely the same time. 

Or are you polling these devices over a limited WAN? In which case i do understand you (but it came to mind after typing the above). 

(CID:93194092)

[DCIM_Support]

The power strips, etc ride on the  IT network, I don't use a separate network just for the devices. Also, they are not all located on my LAN and we have to buy bandwidth to get the data from the device, across the country to the DCE  server.

So I am trying to figure out how to limit that bandwidth usage, to save money (not have to buy a bigger pipe), and limit it to a point where I don't loose data.

Maybe I am asking for the white unicorn.

Regards,

J

(CID:93194118)

[DCIM_Support]

It is possible to configure polling intervals on a per device/group of devices basis.  You could poll the WAN devices much less frequently than the LAN devices if you choose.  This would affect the granularity of your data but not receiving alarms (other than the lost communication alarm managed by DCE, as DCE would only be confirming connectivity the WAN devices based upon your polling cycle).  Adjust this by going to Device - SNMP Device Settings - Device Scan Settings and you can select the devices you want to alter from the global scan settings.  The maximum amount setting for polling is on hour I believe.  Hope this helps

(CID:93194122)

[DCIM_Support]

Jonathan,

Thanks for the reply, however you did mention that doing so would affect the granularity of the data. Does that mean I would only get a data point every hour? or every time the device is polled?

Or would I get the max, min and average of the voltage and current during that polling cycle?

I ask because I want the max amount of current draw which is seen by that power strip so I don't overload it. If the PDU is only providing data whenever it is polled (for that instant) then theoretically I could never see the "max" draw if a device plugged into the PDU is not running during that polling instant.

Maybe that is an extreme but the National Electrical Code calls for '15 minute average' power demand. If I never see the max because of the polling frequency then I could overload the unit. 

This brings me back to my main point, it would be nice to have DCE poll the device and have it send all the data for the time between polls. And more importantly, if the connection between DCE and the device is severed, when the connection is restored, have DCE pull in all the "lost" data because that data is stored in the device itself for a certain amount of days/hours/etc. I shouldn't have to log into a device to try and troubleshoot an event when DCE could do this automatically and it is MUCH easier to compare data in DCE.

Ok, off my soap box. There is SO much more that could be done in this software.

Regards,

J

(CID:93194133)

[DCIM_Support]

John, Alarming isn't affected by the polling cycle (for first party devices), if the device has an alarm then DCE would be notified to poll it immediately independent of the polling cycle. Longer polling cycles mean that your historical data would only have snapshots of the data when it is polled. At any point, you can right click on a device and request the device to be scanned if you would like to see a more recent set of data. Hope this helps

(CID:93913192)

[DCIM_Support]

This question is closed for comments. You're welcome to start a new topic if you have further comments on this issue.