Timed Dispatcher Caching

Sometimes customers want to be more in control regarding the caching of large resources on their Dispatchers. They don’t want the Dispatcher to cache large resources during peak network hours and yet the resources need to be readily available when the Agents/users need the software. This article describes how you can configure a scheduled caching action on your Dispatchers.
Let’s start with a little background information about Dispatcher caching. Resources from your database are cached on your Dispatchers for distribution to your Agent computers. You can configure your Dispatchers to either cache all resources immediately or when they are needed. This can be done in the Global Settings or in the properties of the Dispatcher.

Caching all resources immediately requires no extra administrative effort but you need to be sure that your network is able to take the load. If a large resource is added to the database all Dispatchers will start downloading it to their cache right away, potentially clogging the network. The advantage is that after that every Dispatcher will have the resource available at all times.
If you cache resources when needed nothing will happen if you add a large resource to the database. Only when the first Agent starts executing the job that makes use of the resource the Dispatcher, that is servicing that Agent, will start caching before sending it to the Agent. You can see two disadvantages here:
- That first Agent will have to wait longer because the resource has to be downloaded twice (first to the Dispatcher and then to the Agent) before it can be used.
- You never know when that first Agent will start executing the job exactly. It could well be during peak business hours or during a large backup operation.
On the other hand there is the advantage that not all Dispatchers will start the download all at once and the Dispatchers will only store those resources that are actually used. So in order to make use of this advantage and limit the negative effects of the disadvantages you will have to make sure that the first Agent isn’t bothered by the extra waiting time and the job is executed at a time that you carefully picked yourself.
1. First of all you need to deploy an RES Automation Manager Agent on all of your Dispatcher machines if you haven’t already done that.
2. Then you configure the Agent properties of these Agents so that they have a Dispatcher list that only contains their own computer name or IP address and they only use discovered Dispatchers. This way you ensure that the Agent only uses the Dispatcher running on that same machine.

Make sure that ‘First try autodetect’ is NOT selected.
3. Configure the ‘Dispatcher locations’ and ‘Dispatcher recovery’ settings according to the screenshot below.

4. Set the Dispatcher Cache timing to ‘Cache resources when needed’ and don’t forget the Cache duration. This should be set to several days at least.

5. Whenever you add a resource that you expect to place a considerable load on the network, create a Download resource task that downloads that particular resource to a (temporary) folder on the hard drive followed by a File operations task to delete the newly downloaded resource again.

6. Schedule the module to all the Agents that are running on your Dispatcher machines at a time when the network can easily handle the extra burden. Don’t do them all at once but divide them into batches of 2 or three Agents. In fact you can use a Run Book to let RES Automation Manager help you divide the Agents evenly into batches and make sure that no batch will start until the previous batch has completed. If you want to know how to do that please read my article on rebooting Citrix servers on this blog site.