AWS Load Balancer Settings

[bmarks]

Does anyone here run their Jamf Pro server on Amazon Web Services (on their own, not Jamf's hosted product.) We have been running for two years, but I am doing an audit for performance reasons, and I am wondering if anyone has any recommendations for these two settings:

Load balancer attribute: idle timeout
Two load balancer target group attributes: stickiness and stickiness duration

[bmarks]

For anyone who may come along later asking similar questions, I got some answers:

Idle Timeout: 60 seconds
Stickiness: Enabled
Stickiness Duration: 30 minutes

[Matt_Roy93]

What has your experience been like using AWS overall to host Jamf Pro? Are you running Linux or Windows?  My org is looking at making the switch and was curious about performance and how many servers are actually required?

Thanks!

[bmarks]

Our experience has been very positive. I am hoping to speak at a future JNUC to update my 2016 and 2018 presentations on the topic of optimization. We manage over 60,000 Macs with two Linux-based Tomcat servers behind a load balancer, two memcached nodes and an "Aurora" database server (AWS's MySQL clone that's cheaper, and supported by Jamf.) With that said, I was able to choose resource levels like CPU's and RAM without having to worry about a budget because we're big enough that my team's bill is a small drop in our total AWS expenses. If you have to pick cheaper node-types, this design may not work everyone. We can push out profiles and policies to all 60,000+ Macs at the same time without noticing any noticeable performance drop. We don't manage iPhones with Jamf, so I can't comment on that aspect.

[Matt_Roy93]

Thank you for the info, I am trying to put together an estimate now using their tool AWS Pricing Calculator, we currently have a clustered environment with Windows server VMs and pretty baseline specifications.  With only 3500 computers and very few iOS or iPad OS devices I think we could get away with a dedicated DB instance and one or two JSS instances for everything else.  I am going to push hard for over-estimating on the resources needed just to be safe.

[bmarks]

I'd budget for two EC2's if for no other reason than redundancy, and I'd also pick an Aurora database server for similar reasons because it handles all the redundancy for you (in addition to being cheaper than a comparable MySQL RDS server.)

[Matt_Roy93]

Thank you!