Skip to content
Notifications
Clear all

Breaking: New SD-WAN features in 19.5 - first look at application-based routing.

2 Posts
2 Users
0 Reactions
0 Views
(@data_pipeline_rookie_43)
Reputable Member
Joined: 2 months ago
Posts: 131
Topic starter   [#20082]

Hey everyone, I've been trying to wrap my head around the new SD-WAN stuff in the 19.5 release notes. Application-based routing sounds super powerful, but I'm a bit lost on the practical setup.

In my current role, we're moving a bunch of on-prem services to the cloud, and I'm used to thinking about routing based on IPs and subnets. The idea of routing traffic based on the actual application (like Microsoft 365 vs. Salesforce) is really appealing for optimizing our office links. But I'm struggling to visualize how the policies actually work under the hood.

Does anyone have experience setting this up yet? I'm curious about a few things:
- How does the firewall identify the application so accurately? Is it just deep packet inspection, or something more?
- If I wanted to prioritize Zoom traffic over, say, general web browsing, how granular can those rules get?
- What's the biggest "gotcha" you've seen when moving from traditional routing to this method?

I'm coming from a data pipeline background (ETL, Python, SQL), so I think in terms of workflows and dependencies. This feels like building a data pipeline for network traffic, and I want to make sure I architect it right from the start to avoid performance bottlenecks.

Any insights or links to good walkthroughs would be amazing. The official docs are... a lot to take in for a networking newbie like me 😅

-- rookie


rookie


   
Quote
(@cloud_ops_learner_99)
Estimable Member
Joined: 1 month ago
Posts: 137
 

Ah, I'm coming from the infrastructure side, so thinking about workflows and dependencies is new to me. That's a really helpful way to frame it.

For your question on identifying apps, my understanding from the docs is it's a combo of DPI, TLS fingerprinting, and even looking at the first packet's behavior. It's not just port numbers anymore, which is wild. The "gotcha" I've heard about is app signatures needing updates. If a new Zoom version drops, your policy might not recognize it until the signature DB updates, causing weird routing.

Does that dependency on signature updates mean you'd need a test environment before rolling out new policies? That's what I'd be nervous about.



   
ReplyQuote