Web performance monitors : Tried on a switch, a firewall, an ups. General impression is good, most of it works.

GOOD :
> Nice to see, while in the mib browser, values dynamically changing.
> Counters are correctly recognised, and the relevant math is done.
> Nice to be able to save and re-use the graphs.

BAD :
> If you edit an existing graph to change its scale, the scale does not change.
> Each time you open the mib browser, it goes back to the "mib-2" point. It would be nice to open it back where it was before, instead of doing the whole navigation again.

Web task manager : Tried on windows hosts and a firewall. In general it works also.

GOOD :
> It's really nice to get this type of "familiar" interface for Windows users.

BAD :
> On a windows host, gauges for CPU & PF usage seem to be a bit buggy. The display is garbled at first; after something like one minute, the display becomes correct.

SSG Workspaces : Tried on a bunch of devices.

GOOD :
> It's nice to have this for an ad-hoc, overall look to various parameters of a device or a range of devices.

BAD :
> I'm a little bit concerned that putting many graphs can cause heavy load both on the web server and the monitored devices. I hope at least that polling actually happens only when you show the workspace, not all the time.
> Cpu gauge is weird. I ran it on a bipro Windows machine. Legend says that Cpu1 is green and Cpu2 is red; but the real graph shows black for Cpu1 and transparent for Cpu2. Not a big deal but I still mention it !

Reading, writing and arithmetic - If you need to choose, please take option 1.

Sometimes the non-task manager graphs take a while to reflect what is happening on the remote machine, and they increase or decrease at irregular times.  I ran some tests watching both the SSG graphs, and perfmon on the machine I was testing.  CPU use was only accurately reflected in the task manager SSG's

if you are doing your tests on windows boxes, please pay attention. On these, you can either use snmp or wmi, but it happens that windows does not update snmp information as often as it updates wmi information. Not sure, but this could be the reason why you see "incorrect values".

Most of my tests were on really snmp-oriented devices (firewall, switches, and so on) and the values were accurate when compared to GetIf output.

Reading, writing and arithmetic - If you need to choose, please take option 1.

Thank you for your feedbacks.  The following are answers to some of the issues you've encountered: 

>>If you edit an existing graph to change its scale, the scale does not change.

     I also see this issue.  I will inform the development team.

>>Cpu gauge is weird. I ran it on a bipro Windows machine. Legend says that Cpu1 is green and Cpu2 is red; but the real graph shows black for Cpu1 and transparent for Cpu2. Not a big deal but I still mention it!

     You made a good point here.  The CPU gauge could use a little clarification on it's layout.  The transparent marks are the high and low marks (min/max) for the CPU and the solid dial is the current cpu value.  When looking at the gauge you are only seeing data for 1 CPU instance.  When you click on a CPU in the legend, focus is place on that particular CPU and the gauge will display data respectfully.  The red and green color bars in the legend are there to serve as a visual difference between the multiple CPUs (One could argue that the color bars, in this case, is not needed.  Since the CPU instances has different names.)

>>> I'm a little bit concerned that putting many graphs can cause heavy load both on the web server and the monitored devices. I hope at least that polling actually happens only when you show the workspace, not all the time.

We have been running performance testing to find potential issues in this area and have resovled all issues that were encoutered.  However, if you see any performance issue in this area, please let us know.  Polling only happens when you show the workspaces.

Ipswitch WUG Team,

SV

I wanted to correct my earlier response to the following question:

>>(Web Performance Monitor)If you edit an existing graph to change its scale, the scale does not change.

In my case, I originally had the Web Performance Monitor "Vertical Axis scale" set to "Fixed Scale".  As a result, adjusting individaul counter graph scale will effect how the counter is graphically display but not the "Vertical Axis scale". 

-In the Web Performance Monitor window go to File > Window Properites

-Set the "Vertical Axis Scale" option to "Auto Scale"

You will now see the vertical axis auto scaling with respect to the counter scale setting.

Ipswitch WUG Team,

SV

When you select an snmp or wmi value to graph, you have the option to define the scaling -just like in the windows perfmon. Eg : 1, 0.1, 0.01, 0.001, and so on.

this is what I referred as not working.

Reading, writing and arithmetic - If you need to choose, please take option 1.

Using Internet Explorer 7, Internet Explorer 6 and FireFox 2.0.0.12, I've tested adding multiple performance counters (via WMI and SNMP) with various scale sizes (1, 0.1, 0.01, 0.001, and so on.) to the WUG Web Performance Monitor and did not see any scaling issues.  I also tested scaling counters using a Web Performance Monitor as an added workspace content. 

Can you provide the following information:

-What Web browser was you using?

-Was you using the WUG Performance Monitor Tool or was it an added content in a workspace?

-How many Performance Counters are in your WUG Perfomance Monitor?

-What Window Operating System did you install WUG on?

-Specifically what performance counter scale sizes did you try?

Ipswitch WUG Team,

SV

IE7, latest patches.

-Was you using the WUG Performance Monitor Tool or was it an added content in a workspace?

Performance monitor tool.

-How many Performance Counters are in your WUG Perfomance Monitor?

At that time, there were only two snmp counters. One was Cpu usage and the other memory usage. Since Cpu range is 0-100, it would appear completely "flat" so I decided to change the scale for memory usage from 1 to 0,001. But it did not change.

-What Window Operating System did you install WUG on?

Windows 2003 R2, SP2, latest patches.

-Specifically what performance counter scale sizes did you try?

0,001 - 0,01 - 0,1 - 1

I will check tomorrow if I can reproduce it and will let you know.

Reading, writing and arithmetic - If you need to choose, please take option 1.

A team member notice your syntax when typing the scale value.  You used commas instead of dots, for example 0,01 instead of 0.01.  This suggested that you may have WUG installed on a foriegn Windows OS instead of an English Windows OS. 

As a result I retested using a German Windows XP operating system and found that the performance counter scale issue does exist when WUG is installed on a German Windows OS system.  Thank you for bringing this issue to our attention. 

Ipswitch WUG Team,

SV