HOWTO: Calculate HPDM Gateway polling interval
To make sure the console reflects the correct Thin Client status (Active / Inactive) the Gateway van poll the Thin Clients to see if they're still active (or not). Therefor it uses an polling interval. But how do you calculate this value?
By default the HPDM Gateway does NOT poll the Thin Clients. In the Gateway configuration settings the polling interval is set to 0. To start the poll process this value needs to be equal or bigger than 60 seconds (the minimum).
But which value is valid for your environment? Well, this completely depends on the amount of Thin Clients and Gateways in your environment.
How does polling work?
Polling consists of two settings:
- Polling Batch: The amount of Thin Clients which are polled at once (range from 1 to 50). Sending out the poll signal to the batch takes about 1 second
- Polling Interval: The time (in seconds) between the polling process.
Then:
- The Gateway takes the size of the Polling Batch and sends out a poll signal
- The Gateway takes in alphabetic order of the HPDM ID (aka the Thin Clients MAC-address) of the Thin Clients the IP address.
- The Gateway tries to create a connection on the IP-address to the HPDM agent.
- When the connection is created (in other words; the IP address is still alive and this device has a HPDM agent installed) the HPDM ID in the database is compared with the actual one from the agent.
- If no connection can be made or the ID is different the device is marked as Inactive.
- Otherwise it is marked as Active.
- Gateways waits 30 seconds and polls the next Polling Batch
- The Gateway repeats this until he reaches the end of the connected Thin Clients
- Then the Gateway waits for the Polling Interval passes
- And restarts the whole again
How do you calculate the Polling Interval value:
- An example:
- 500 Thin Clients
- 1 Gateway
- Polling batch : 50
- Polling Interval : ?
500 TC’s / 50 TC’s per batch = 10 batches
(10 Batches – 1) * 30 seconds = 270 secondes
10 batches * 1 second per batch for sending out the polling signal = 10 seconds
270 + 10 = 280 seconds.
So the polling interval should NEVER be lower then 280 seconds in this case.
To make sure the gateway has ‘space’ enough to do his job and not to overload the network with polling signals, we suggest to put the Polling interval to 900 (once per 15 minutes).
Again, polling is just a mechanism to make sure the console reflects the correct status of the Thin Clients. When your environment consists of desktop Thin Clients and users have the ability to shutdown the Thin Client on a normal way, the chance that the status of the Thin Clients in the console does not reflect the true status is very small.
But when you environment consist of a lot of mobile Thin Clients, and the users are used to put the TC into sleep by closing the lid (so the HPDM agent does not have the chance to report his closing status) then the chance of not reflecting the true status is very big.
Normally a polling interval of 7200 seconds (2 hours) should be enough for most organizations. But if you want to make sure the console reflects quicker the true live status, you can use the above calculation for the polling interval.
Where do I configure these polling settings?
In the Console got to:
- Gateways & Repositories
- Choose on the left-hand side for ‘Gateways’
- Select the Gateway
- Right Click with mouse and select ‘Configure HPDM Gateway’
- Enable the two options, leave the others disabled.
- Configure settings
- And press OK to start the configuration of the Gateway.
- Repeat these steps for all the gateways in your environment.
What about more Gateways?
If you have more than one Gateway you need to know how many Thin Clients are connected to each gateway and repeat the above calculation per Gateway and set the correct value per Gateway.
Do I need to do this also for my DMZ based Gateway?
There is no need to enable polling on Gateways which are setup in DMZ to manage internet connected Thin Clients. Internet connected Thin Clients are mostly situated behind NAT-firewalls (at home or at Internet connected workspaces) and 99,999999% (and maybe we forgot some nines behind the comma 😀) of them will block the polling signal. So do not disturb you DMZ based Gateway with the polling process and set the polling interval for this one to 0.
What if I set the polling interval too small?
When the polling interval is set too small, the Gateways gets in a polling loop and gets overloaded (he keeps polling and got no time for other tasks). This is noticeable by the ‘Broken’ status of the Thin Clients in the console. So avoid in all times to set the interval smaller than calculated.