Load Balancer for 'Outgoing' Service?

[WaiKeat]

All these while, the Load-Balancer implementation or use-case I've heard of (and also read in documentation) is for Load-Balancer front-ed the servers and/or service listening for incoming traffic and/or request. For example, any website that serving the customer nation-wide or worldwide.

I am wondering how about service that actually has outgoing traffic instead of listening for incoming traffic/request? For my case, I would actually spawn multiple VMs in GCP to act as load-generator that pump load to my customer for their Stress-Test project.

So instead of spawn multiple VM and give multiple public IPs for my customer to whitelist them, I want to put one load-balancer front-ing all these VMs so that I only need to pass one public IP for customer to whitelist (and also, in-case I need to spawn more VM behind the Load-Balancer, I can do so without problem since I don't need to bother customer to whitelist additional IP)

Can I configure load-balancer in such a scenario? Scenario where load-balancer is expect to 'mask'/'protect' all the IP behind a client-side, instead of the usual scenario which is masking all servers IP behind server-side.

And if yes, any configuration steps that looks significantly different from the usual scenario?

Best Regards,

Wai Keat

[glen_yu · Accepted solution]

I think what you're looking for is a Cloud NAT .  You can provision a bunch of compute engines without public IPs and just have them use the Cloud NAT as the outgoing/external IP.

Better yet, if you can generate the node via Cloud Run or App Engine, you can configure Serverless VPC Access and use Cloud NAT there (this method will require more setup though, so if you just want something quick and dirty, the compute engine way is easier)

[WaiKeat]

Thanks Glen.