This message was deleted.

Using AppFlow/HDX Insight v1 on a VPX 200 is going to be a bad time for anything more than a handful of users. You've got only one CPU core that's doing encryption, packet handling, appflow parsing, etc. It's highly dependent on the average throughput per user, core speed of the underlying host as well as contention and whether it's hyperthreaded or not.

and SSL key length too, that will make a massive difference

Atleast make sure that your using logstream over ipfix as the transport for appflow, I’ve seen huge improvement by switching over.

Definitely. And since the new LTSR includes the necessary components for NSAP/HDX Insight 2.0, once people start testing/implementing that the processing load for Insight on NS goes way down.

So, the golden question, any documentation or ways to verify that this is the issue? lol

Uploading a bundle to CIS should show you if there are queuing/contention/etc issues

https://docs.citrix.com/en-us/netscaler-mas/12/analytics/logstream.html AppFlow with ipfix has been an issue for long while, with crashes, reboots and bad performance. What the issue with your customer is, is hard to say without knowing more, but if i should take a guess it’s running 11.1 or if newer it’s still using ipfix.

They are using Logstream (just verified), and I am working on getting them to pull a support bundle now. Thanks for all the help as always guys

Which firmware? How many users are accessing through the ADC? What kind of workload? Is the hypervisor presses for resources?

12.1 55.18, @100 or so users all ICA connections, Trying to get some info on the hosts.

You could always assign more cores to the VPX, you aren't limited to what's in the OVF.

The license level caps the packet engine instances which links to the number of cores and ram so that wont do anything unless the license is above a vpx200

oh, didn't realize that

I'm just trying to find the ncore article that explains the sizing. It's not very well documented. Essentially for each additional core you add a certain amount of ram and you'll get an additional packet engine.

Are you saying that’s possible even with the VPX200 license? I would think not, lol. We are working on gathering data to support the upgrade of the license. We uploaded a support bundle to CIS, only thing it found worth mentioning was that invalid http requests aren’t dropped, but that is being corrected this week

@Ryan Chaney have a read of the article above, basically no. You'll receive no additional benefit from adding more resources on a vpx200.

I'm sure I saw a Citrix blog or article several years ago showing 1500+ concurrent users through a VPX200, but can't find it now

Maybe with synthetic benchmarks and not actual user sessions, but from the latest numbers I've seen the VPX200 is rated for around 600 concurrent task user type sessions without HDX Insight.

Yes I've certainly never seen anything near those numbers in production