Jump to content
Welcome to our new Citrix community!

Netscaler VPX200 Throughput monitoring


Recommended Posts

Hi everybody,

 

I know that this kind of question has been posted many times, but I can't find some confirmations on my thoughts.

 

I've got a NS VPX200 (v11.1.58.13) and want to be sure that I'm right on my monitoring.

During last French strike ;), we faced more remote connections than usually and got a lot of complaints from users, as the sessions were not as smooth as usual.

(no heavy load on the XenApp servers, as users instead of being connected internally, used an external access point through the Netscaler, and we have an average load on XA servers of 50% Max).

 

Network link above my NS was only 100Mbps, so half of the NS capacity.

Here is the graph I used to show that my NS was (obviously) not in cause --> Max bandwidth is only around 10Mbps, 

 

 

Questions : 

 

1- Does it mean that I'm only at 5% of the max bandwidth capacity of the NS (10/200Mbps) ?

   Think yes, but network team tell me that their link (100Mbps) was not fully loaded, so why not more than 10Mbps used ? and why slow connection for end users ?

 

2- During real time monitoring, I saw a long period where the peak was around 13Mbps, but can't find any trace of it, even by selecting the exact hour, max visible now is only 10Mbps...

   How can I find it again ?!

 

3- With around 400 sessions, for 10Mbps, it's only 3kB / session... when Citrix or guides speak about 30 to 250kB/s depending of the usage.

What 3 means ! my users were sleeping all the time ?!

(note on the AMC, at this time, maybe 50 users were idle, but not 400 !)

 

4- If anyone can point me to a good monitoring guide to understand all monitoring counters or give me the list of useful one (main usage of the NS is ica connection from remote access).

 

5- I can't found a formula to calculate the bandwidth needed regarding the screen resolution, nb of monitors, type of usage,...  if someone have something !

 

Thanks for reading this long post, 

 

Nicolas M.

2019-09-24 10_02_47-Citrix NetScaler VPX - Reporting.jpg

Link to comment
Share on other sites

Your issue may not have been one of bandwidth utilization necessarily. A number of things can influence session performance via Gateway, especially things like the TCP profile bound to the vServer. Do you have a TCP profile bound? The default XA_XD profile usually needs to be cloned and tuned a fair degree to work well. HDX Insight data via Citrix ADM is particularly useful in more accurately determining session bandwidth as well. Session bandwidth utilization can vary significantly. If users are using basic application with limited moving imagery, little to no client redirection, and frequent bouts of idleness, traffic will definitely be lower. Minimal traffic will traverse the wire on idle sessions.

 

Another thing to look at... are you using VMware as a hypervisor? If so what NIC model is used in your deployment? E1000 or VMXNET3? E1000 was not supported in many firmware versions, but more current versions of 12.1 brought back support.

 

On the topic of firmware, I should point out you're a year behind, and the version you're running on is subject to at least one security vulnerability (https://support.citrix.com/article/CTX240139) that warrants your VPX platform being patched as soon as you're able, either to the latest 11.1 firmware or the latest 12.1 maintenance release of 54.13, if permissible and subject of course to backup and testing.

Link to comment
Share on other sites

Hi Michael,

 

No TCP profile bound on my vserver, only the http is bound.

I’ll have a deeper look at this part of the conf as on other Netscalers for other usages, it is set.

 

Ok for HDX Insight, I’ll try to found some time to deploy it as many new branch offices are going to connect to our infra.

Seems quite useful.

 

Yes we are on VMWare, using VMXNET3 as E1000 uses too many cpu.

 

Ok for the update, daily prod consumes all our time, but yeah, I’ll plan the update to the 12.1 (backup+test included) !

 

Thanks for your answer,

Nicolas.

Link to comment
Share on other sites

  • 2 weeks later...

Take a look at https://support.citrix.com/article/CTX207005 , which explains that, for SSL traffic, the TCP settings are not aggressive enough, and describe what to do....I strongly suggest you take the steps listed.

 

The problem with graphs like the one you show is that they tell you lies: on AVERAGE you probably have enough bandwidth, but every so often, for a few seconds at a time, your link will be saturated, causing momentary slow-downs. With a graph like that, the 10 meg will be an average over many minutes! If you use the dashboard (when under load) then you'll get slightly better stats, as the figures are averaged over 7 seconds.

 

And now for the silly question: are you 100% sure that your ethernet is actually running at 100meg? Or gigabit?

 

And what do your users mean by "slow"? Compared to other "quieter" days when the connect remotely? or compared to when they are in the office?  If the latter, then yes, it will be slower: I'll bet in the office, where they connect directly to the XA/XD machines, and they are using gigabit, with almost zero latency... rather than via a 100meg link with MUCH higher latency! And it doesn't need a lot of traffic to increase the latency.

 

This is of course where ADM (aka MAS) could help, with it's ability to provide "HDX Insight", which will give you REAL performance data (= latency data) that relates to the "slowness" of a session.

Link to comment
Share on other sites

Tests are currently running with the TCP profile. Looks good.

 

Regarding graphs, I was always wandering what is the interval of measuring...  Thanks for the 7s ! makes sense about that it's not as accurate as I think !

 

Yes, it was a 100Mbps link, upgraded since.

It was really slow, I was impacted as well, so for once, it was not some 'Citrix is bad' common users statement.

Frequent freeze (1 to 10sec),  disconnections too (maybe 10/hour)... really a bad day to work remotely

 

I raise the MAS need to my boss, hope we can find time for that, as it looks really good.

 

Thanks you both for replying.

 

Note on point 3:  Citrix sessions were at 3kB so around 24kb, which is not enough, but not as dramatic as I thank before!

 

 

 

 

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...