search

F5Networks / f5-azure-arm-templates-v2

Azure Resource Manager Templates for quickly deploying BIG-IP services in Azure
22 stars 45 forks source link

Can I Auto-scale f5 VE with 1 instance as min #42

Open aymanelbacha opened 2 weeks ago

aymanelbacha commented 2 weeks ago

I need confirmation since there's no reference if Auto-scale works for 1 instance only as minimum

mikeshimkus commented 2 weeks ago

Hi @aymanelbacha, the minimum number of BIG-IP instances in the autoscale templates is 1, per the bigIpScalingMinSize parameter, which defaults to 1: https://github.com/F5Networks/f5-azure-arm-templates-v2/blob/48fbe5f9b4262e88ad37c98b6aae27cc0591d9e3/examples/autoscale/payg/azuredeploy.json#L114

Let us know if you have issues using 1 as the minimum.

aymanelbacha commented 1 week ago

Thank you @mikeshimkus for the response, no issues so far, but checking the options we have. Meanwhile, I don't see any dedicated HA sync interface/VPC to sync between instances in the existing repo, are we using the management to sync, is it better to have a separate NIC, how it behaves when autoscaled

appreciate your support

mikeshimkus commented 1 week ago

Hi @aymanelbacha, check out the README here: https://github.com/F5Networks/f5-azure-arm-templates-v2/tree/48fbe5f9b4262e88ad37c98b6aae27cc0591d9e3/examples/autoscale/payg#changing-the-big-ip-deployment

There is no synchronization between instances in the cluster. This solution uses an immutable model; the configuration is updated by modifying the Runtime Init configuration and redeploying the template while changing the value of the bigIpRuntimeInitConfig parameter to use the new config URL. All the instances are then redeployed using the new configuration.

aymanelbacha commented 1 week ago

Hi @mikeshimkus , I am working on a GCP not Azure, is that something supported, how to overcome this, since there will be daily changes, for example create new Virtual Servers/nodes... would it be like have a automated solution that connects to both instances and future auto-scaled and push these objects without redeploying?

mikeshimkus commented 1 week ago

The same model applies in GCP. The solution as-is does not support updating configuration directly on instances, as this configuration will not be applied when new instances are added to the autoscale group. However, it should be possible to create a Google cloud function to check for scale out and apply configuration to new instances. This is not in scope for our templates.