Here are a few queries from the Forward Library that perform math (percent calculations) as examples.

FQ_fe5c9a8ddc1fc26236a0979fc4d085c552aea917

FQ_916976063d3cd8adde94d4afc178724fcc2ee088

FQ_4de6e4a4af3c8e4dce1d7bd1e446d95752d30ec8

Notice that I multiply by 100 first before dividing.  At the time we only did Integer math.  There now is the ability to set a Type of Float, if you actually want decimal places.

I’ll look over your post a bit more soon.

-Tyson

There is some examples of maths also in the nqe documentation.

However, I think you may fiind it be better to consider the ports not used as a number and avoid the percentages. Percentages need to be carefully examined as it can hide the truth.

For example 50% used sounds like we have too much spare capacity, but this may only be true if we actually have a fully populated chassis with many cards installed. On a 24 port switch it sounds okay at least to me. If a switxh has only just been installed then we would expect it to be under uaed for a period of time.

If you do this as a percentage then you may want to first group similar devices. Those that are for the same site, environment or what ever makes sense for your team. As the percentage is then across mutiple device then i think iy may sit better in any reports.

You may also want to flag what the port is used for, such as networking, servers, users, printers and so. Perhaps the descriptions on interfaces can help or whether the port is enabled may be important. If i recall i think NQE exposes earned hosts as well which may be incorporated into your reporting.

For example saying we have 4000 ports for servers available and 2000 are in use for location X gives us 50% utilised is probably more useful for planning.

This isn't to say that having the specifics for each device isn't needed as well.

To group devices you would need some method such as tagNames, naming convention, etc  to grab the collate the data for related devices.

@cariddir  This is purposely verbose so you can see the logic. I have two functions which test for any 0 counters or all 0 counters


 

test =
  """
Router# show interfaces counters
Port            InOctets    InUcastPkts    InMcastPkts    InBcastPkts   InDiscards
Fa0/0        1234567890         123456         123456          12345           0
Fa0/1        9876543210         654321         654321          54321           0
Fa0/2                 0              0         654321          54321           0
Fa0/3        9876543210         654321         654321              0           0
Fa0/4        9876543210         654321              0          54321           0
Fa0/5                 0              0              0              0           0
""";

pattern =
  ```
{Port:string} {InOctets:number} {InUcastPkts:number} {InMcastPkts:number} {InBcastPkts:number}
```;

countInactiveAnyZero(interfaces) =
  foreach interface in interfaces
  where 0 in interface.counters
  select interface;

countInactiveAllZero(interfaces) =
  foreach interface in interfaces
  where sum(interface.counters) == 0
  select interface;

foreach x in n0]
let blocks = parseConfigBlocks(OS.UNKNOWN, test)
let matches = blockMatches(blocks, pattern)
let fixCounters = (foreach match in matches
                   select {
                     port: match.data.Port,
                     counters: :match.data.InOctets,
                                match.data.InUcastPkts,
                                match.data.InMcastPkts,
                                match.data.InBcastPkts
                               ]
                   })
let zeroInterfaces = countInactiveAllZero(fixCounters)
where isPresent(zeroInterfaces)
let activePorts = float(length(matches) - length(zeroInterfaces))
let totalPorts = float(length(matches))
let percentage = activePorts / totalPorts
select {
  zeroInterfaces, totalPorts, activePorts, percentage: percentage * float(100)
}

Gary,

Thank you very much. 

 Going to work with this today.  I’ll let you know how it works out!

Again, much appreciated.