Toma,

you wrote:

BELITCHKI, Toma (HO) wrote:

Hi Guys,
I would like to ask for your experience in regards to the Server Sensors
update process.
Do you consider it more like an anti-virus update - wherever you have new
attack definitions, you apply them automatically, or you first test the
update on test machines, perform degradation testing, etc. and only if the
tests are conclusive - update the production machines.

I would definitely agree with your first conservative approach. Testing things first before you break you production servers is always a must. Of course this should not delay the overall update process too mutch. To speed up the regression test process, we have a set of inhouse machines to run sanity checks on new software. In larger customer environments, a server sensor test lab is always ap and running to have the ability to test the compatibility of new service releases and XPUs.



We run our sensors on a mission critical machines and it's unacceptable to
brake the server's functionality because of the Server Sensor's update. That
implies thorough testing of the updates before going in production, from
other side, this is resource and time intensive job which delays the Sensor
update process.

Right. The classical management principle is security over functionality. However a server that does stop working is just as insecure as one that isnt patched. You may ask why that is. Think of an ADS that takes care of the security policy in your network and isnt working, because of an update. If its brought down, administrators and users may have to workaround its functionality, e.g. using local accounts on their machines that are not audited. Thus your security infrastructure is bypassed and you have made things even worse.




I would like to hear about your experience, have you ever had a problem, where applying the XPUs would break the server? I personally used to look to the Server Sensors as non-intrusive components of the system, until we had the problem with the rsiisfilter.dll ISAPI filter, who was causing the inetinfo process of our web server to go 100% CPU (it was fixed with SS 6.5 SR 3.4).

This was a real exceptional scenario and we have not experienced this, because of the defensive update policy described above. In general a server sensor update will not break your system, but you dont know for sure. So testing is still required. From my experience the implicit risk of an XPU is much less, than a service release or major release upgrade.
If there is a problem it will typically affect the sensor functionality, not the server functionality. Still you should make sure, that you have a backup strategy in place that allows for rolling back the update quickly, e.g. by using engine mgr procedures that can be automatically run in the background.



Just my 0.2 cent Karl



Regards,


Toma
_______________________________________________
ISSForum mailing list
[EMAIL PROTECTED]

TO UNSUBSCRIBE OR CHANGE YOUR SUBSCRIPTION, go to https://atla-mm1.iss.net/mailman/listinfo




--
______________________________________________________

BDG GmbH & Co. KG - Make IT safe.
Wendelinstr. 1
D-50933 Koeln

Tel:     +49 (0)221-954231-0
direkt:  +49 (0)6126-94433-21
Fax:     +49 (0)6126-94433-31

E-Mail:  [EMAIL PROTECTED]
Web:     www.bdg.de
______________________________________________________




_______________________________________________ ISSForum mailing list [EMAIL PROTECTED]

TO UNSUBSCRIBE OR CHANGE YOUR SUBSCRIPTION, go to https://atla-mm1.iss.net/mailman/listinfo

Reply via email to