How to monitor bandwidth limit for application control

[Kosin_Usuwanthi]

Hi

Anyone can advice. How to monitor bandwidth limit for application control ?

MGMT R80.10

GW R77.30

[Danny]

SmartEvent Views is what you are looking for.

[Kosin_Usuwanthi]

Hi Danny

Thank you for your advice but I would like to see real time utilize or exceed drop for limit policy  when user complain about slow access.

[Nico_Seddig]

Hi all,

after a long time for that question i would like to ask if someone found a real solution . I have the same question and tried to figure out the possibility with a view from Logs&Monitor page. The problem ist that  i found based on sk144192 the table fields for limit_request, limit_applied, etc. but i cant choose it within the table settings of the view. The table fields were indexed but no name that i can search for. Any hint would be helpfully.

[PhoneBoy]

SmartEvent can only create reports on fields that are directly indexed.
These fields are not, presumably, which is why they can't be specified for reports.
Sometimes, specific fields can be enabled through an RFE with your local Check Point office.

[Nico_Seddig]

Hello again, thanks for the quick response. I will go back to our customer with that information and will discuss it maybe with our Channel Partner SE. Thanks and have a nice day.

[Kosin_Usuwanthi]

Note - The Security Gateway implements the Limit action by dropping successive packets which exceed the allowed bandwidth.

How to monitor exceed drop ? I have no see this log from SmartConsole.

[Danny]

Then you want to use SmartLog like this:

[Ed_Eades]

I do not get any results when using the filter, "bandwidth AND appi_name:YouTube".  However I do get results if use the filter, "bandwidth" by itself or "appi_name:YouTube" by itself.

I am very interested in getting the results like the screen shot from Kosin.

[Hugo_vd_Kooij]

Bandwidth management by dropping packets is a bad strategy. You can get much better results by changing (reducing) window sizes in responses and holding back acknowledgement packets.

Nothing gets dropped but you let the TCP protocol do the bandwidth limiting.

Not somethink I invented. Companies like PacketShaper (bought by Blue Coat, bought in turn by Symantec) did create great appliances with features like this.

[Kosin_Usuwanthi]

Thank you Danny. Now I can see drop log from smartview for bandwidth limit policy.

[ovidiu_catrina]

hi

i have the same question as you.

Could you please guide me on how to get those statistics as you did?

thanks

[Greg_Harbers]

Hi,
are you able to provide any additonal details as to how you got this report?
Thanks

[Timothy_Hall]

When researching my book I dug into the APCL Limit feature, trying to find a way to disable APCL limits "on the fly" for testing purposes or to monitor real-time statistics for packets dropped due to a Limit.  The goal was to ensure an enforced Limit was not the cause of poor performance.  Other than the statistics that are included in the traffic logs as demonstrated by Danny Jung‌ above, there isn't a direct way to do that.  I also discovered that APCL/URLF cannot be disabled "on the fly" like Threat Prevention can with fw amw unload.

However fw ctl zdebug drop will show real-time packet drops due to an APCL limit with the message: PSL Drop: APPI_LIMIT. 

Also watch out for this issue when limits are applied to traffic subject to HTTPS Inspection:

sk70600: Connectivity issues when configuring Application Control limit and enabling HTTPS Inspectio...

--
Second Edition of my "Max Power" Firewall Book
Now Available at http://www.maxpowerfirewalls.com

[phlrnnr]

@Timothy_Hall doesn't sk70600 only apply to R75.40?  Or does it apply to the R80 trains as well?

[Timothy_Hall]

The cpas_tcp_do_sack variable still exists in the R80.40 kernel and its default value is 1.  If I'm reading the SK correctly you only need to flip this variable if the two hosts are using very large TCP windows and packet loss is present.  There were a ton of changes to CPAS (R80.20) and TLS Inspection (R80.30) so this variable may not even be relevant anymore.