OLD (APM 3.5 now GA) - If you're curious as to what we're working on...

The APM team is busy at work on a number of highly requested enhancements:

Filtered Application-specific Summary Views
- Ability to create a top-level summary view that is filtered to show data from a specific application
- E.g. Exchange, SQL, IIS, Active Directory, etc.
Shareable Application Dashboards
- Ability to create an application-specific details view and associate that to an application template. For example, you can create an Exchange Application Details View that shows a different set of chart and graph resources than your SQL Server Application Details View. This is analogous to what you can already do today in NPM where each Device Type can have a specific Node Details View.
- Ability to add specific component monitor graphs and charts to your custom Application Details View. For example, on your Exchange Application Details View, you can add a custom area graph or chart for a specific Exchange process, eliminating the need to drill down to the monitor details view to see critical data
- Ability to share and import dashboards by simply exporting or importing the application template.
More built-in Application Templates, including Exchange 2010
Enhanced Performance

In parallel, we're also working on the following longer-term items:

Groups / "Containers" (name TBD)
- Ability to create select multiple objects even of different types (e.g. nodes, interfaces, applications) and group them into a container that can be used to visualize status and for configuring "service level" alerts
- Ability to see "containers" at the summary and details view levels
- Ability to set group level status as worst (worst status rolls up), mixed (yellow unless all objects are green, or all or red), or best (best status rolls up for high avail/hot backup scenarios)

PLEASE NOTE: We are working on these items based on this priority order, but this is NOT a commitment that all of these enhancements will make the next release. We are working on a number of other smaller features in parallel. If you have comments or questions on any of these items (e.g. how would it work?) or would like to be included in a preview demo, please let us know!

Find more posts tagged with

trigger

applicatione_template

unmanage_application

Component

text

SNMP

apm

assign

component_monitor

script

WMI

feature

credentials

Accepted answers

All comments

FormerMember

unfortunately the only feature i would like to see in APM in the next release, is for the ability to be able to select if you to collect any performance stats or not on a process, of if you ONLY want to know if a process is running one or more times... if i can not get this feature, then APM is unfortunately going to need to be shelved for me, which is bad news as I have ALX, and 4 additional pollers with APM poller component. APM is killing my NPM database due to it collecting performance data on thousands of processes i am just not interested it

any thoughts? it can not be hard surely to add this?

chris.lapoint

ability to be able to select if you to collect any performance stats or not on a process, of if you ONLY want to know if a process is running one or more times

I'd like to make sure I've captured this requirement correctly and ensure we don't have any workarounds that we could deliver out of band.

Can you please describe in more detail the use-case and the problem you're trying to solve with this feature?

FormerMember

Hey Chris!

I hope you are well!. The problem i have, is that my lovely NPM installation is being hammered by APM. Basically, i want to use APM to extend the availabiloity monitoring of Orion, to be able to tell me when specific processes are UP or DOWN. I rarely care about CPU, or Memory statistics on processes, i just care whether there is one running or not.

I believe APM causes major performance issues for me due to the amount of processes that are discovered when APM polls for its existence.

Here is an example:

- I create an application template, with one SNMP component.

- The SNMP process name is “cron”

- I assign the application process to 10 Solaris servers

- Each Solaris server runs 15 instances of “cron”

- All i care about is knowing if there is at least 1 “cron” process running on each server, and i do not care how much RAM or CPU they are using

- APM discovers all 15 processes, on all 10 servers, and collects stats on each one and writes them all to the history tables.

Imagine that scenario, but I have 300 servers all running multiple instances of processes where I only care if they are UP or DOWN. It means I am recording thousands upon thousands of database read/writes for information i just do not care about. 2 of my APM tables went up to 80GB last week, but the reast of the NPM was only 10GB, and that is around 10-12000 elements.

If there was a way when I create an application template to say “intelligent” or “simple” etc, where literally the “simple” template just counts the number of instances of the names process, then i would be set.

I understand this is not as trivial as i describe it, but there is at least no new technology required as you already have a variable for the count of the number of processes.

What do you think?

jiri.tomek

Hello Dan,
I posted reply to (your previous thread about this issue) to keep things at one place.

chris.lapoint

I understand this is not as trivial as i describe it, but there is at least no new technology required as you already have a variable for the count of the number of processes.

What do you think?

It seems straightforward. Do you need to be able to know which specific processes are up/down or do you simply care that 10 instances of a specific process are running on a given server?

FormerMember

How about this.. A place to put notes on a node.. a interface.. an application service a whatever. Not all objects need comments.. but there are always those one offs that would benefit from having special instructions.. support contact info.. notes? just a thought

anachronism

How about this.. A place to put notes on a node.. a interface.. an application service a whatever. Not all objects need comments.. but there are always those one offs that would benefit from having special instructions.. support contact info.. notes? just a thought

We use custom properties for this, but that doesn't work (I don't think) for applications, just nodes, interfaces and volumes. If you have a bit of time on your hands, you can also do some cool stuff with a custom HTML panel, some javascript and a wiki to hook in all your nodes (but that's a little bit more work... Something like this might work though in your custom html):

-------8<----------8<-------

document.getElementById("Wiki").src="">wiki.mycompany.com/.../view + document.getElementById("ctl00_ctl00_ContentPlaceHolder1_ViewPageTitle_NodeLink1_lnkNode").innerHTML;

</script>

-------8<----------8<-------

But obviously you'll have to customise the url for your own wiki flavour. Adding in a height tag on the iframe would be good too.

Stealthn

I'll second the need for application groups. If you want to be an Enterprise application monitoring tool you need to be able to monitor an application as a whole, from the presentation server, to the app server to the database, and there might be many of each type of server.

There are not too many Enterprises that are runnig key business apps on only one server, and this realyl is the only space for the current version of APM. IMHO

Bob James

Miron

Hi,

I thought I would add to my feature requests:

Simple URL Check.

At the moment when you want to add a URL to monitor you have to create a new Application Template and assign that tempate to a node. Now while this is not to bad It would be simpler to for instance just as you add a node, you select add URL. This would then bring you up the elements of the template and instead of asking which node you want to assign the template to it does a lookup and adds it to a node if the node is existing,or creates the new node for the user.

You could have two options:

1) when you click URL it just brings up a URL input box and when you click submit it adds to existing node or creates new node.

2) when you click on URL it brings up URL input box with an option to select advanced, if user selects advanced bring up whole applicate template, and then follows addition to node in similar fashion to above

I see this being used by the web developers or content providers that can create the content then quickly add it to Orion for monitoring.

Miron

dayley

Here is a partial list of our APM/NPM frustrations / feature requests:

1) Truncation of returned component messages to 150 characters. Orion has the ability in Windows script monitors to return message text to Orion. We pass it along to our trouble ticketing system, but it arrives incomplete because Orion truncates the text. See:

2) Limited ability to pass Orion variables to Windows script monitors as script arguments. Only ${IP}, ${USER}, and ${PASSWORD} are supported (they're called "tokens). The ${SQL: ...} query format doesn't work in the argument field either. Even if it did, the field in the web interface is too narrow to support any but the simplest queries. We can't even pass the computer name ("CAPTION" in SW parlance) to a Windows script. To get around this in our scripts, we have had to embed a database connection/query look it up. NOT good. See: http://www.solarwinds.net/NetPerfMon/SolarWinds/wwhelp/wwhimpl/js/html/wwhelp.htm#href=OrionAPMAGCreatingWindowsScriptMonitor.htm and

3) No ability to enable/disable specific alert trigger actions. This is ameliorated by the ability to export/import actions, but it would be very useful to be able to toggle them instead, especially during new alert testing.

4) In the APM Templates view, when creating a new template, it would REALLY be nice to be able to select the tags to assign from a list. This THWACK forum has something like this capability. Can it be ported to APM?

5) Also when creating a new component in a template, it would REALLY be helpful to have a "Copy" button to insert a duplicate component that would then only need to be modified slightly instead of having to be entirely created from scratch.

6) In the "Manage Assigned Application Monitors" view, on the "Assigned Application Monitors" paging is limited to 25 assigned application monitors. This gets tedious when one has to look through > 50 pages of these. It would REALLY be appreciated if the number of monitors displayed per page is a configurable option in the view. This interface is so wearisome that when we have needed to delete a number of monitors, we found it much easier to just make a copy of the template and delete the original, and then reassign the monitors. That wipes out all historical data, but there have been times when it was worth it.

7) If ever there is a problem with one of the monitor templates (e.g. it pegs the poller's or nodes' CPU usage), it would be REALLY advantageous to be able to immediately disable the template (and consequently all of its components).

8) On 64-bit Orion pollers, the "Execute and external VBScript" alert trigger action needs to be able to run as either a 32-bit or 64-bit process. If Orion is using the cscript.exe interpreter for this action, then on 64-bit pollers, it should use the 32-bit version of cscript.exe (usually found in C:\WINDOWS\SysWOW64), instead of the default 64-bit version of cscript (usually found in C:\windows\System32). We have been able to get around this on our 64-bit pollers by having the alert trigger run a 64-bit cscript.exe interpreted script that, in turn, runs a script with a 32-bit cscript.exe interpreter. This is clunky. A much better solution would be for Orion to expand its list of VB Script Interpreters to include both 32-bit and 64-bit types.

9) In the "Edit Execute VB Script Action" panel, there needs to be a "run as" configurable option so that the scripts that are executed can run either as the account associated with the SolarWinds Alerting Engine service, or as the credential assigned to the monitor. The lack of this capability has been a significant problem when running scripts that require different credentials for different environments. Incidentally, it would also be handy to have an "Insert Variable" button that pops up the "Select Variable" panel (just like in the "Edit E-mail/Page Action" panel) and inserts the selected variable into the "VB Script to execute" field.

10) Handling forwarded SNMP traps from an external source. In the Trap Viewer Settings, on the Alerts / Filter Rules, when a new trap rule is created or edited, there is no action available in the Alert Actions tab to directly create an Orion APM alert (just options such as Tag the Trap, Forward the Trap, Send an E-mail/Page, etc.). To get around this, we have configured the traps to go to the poller event log, and then set up an APM monitor to pull them from there. This is inefficient.

11) We would like the option to use a full SQL Server version for the NPM/APM job engine instead of the SQL Server Compact Edition file (in C:\Documents and Settings\All Users\Application Data\SolarWinds\JobEngine\Data\JobEngine35.sdf) for scheduling. We understand that this is an option that not all SolarWinds customers would require, but we have serious reservations about the performance and size limitations of the Compact Edition. See:

12) In the Account Manager, it would be beneficial to include a Yes/No drop-down option to "Allow Account to Un-manage Assigned Applications" so that select users can have the ability to do this without having to have Administrator rights.

We have had other problems with Orion for which we have tickets open to address them specifically. Please note: The above list is intended as a starting point for SW for improvements in the next version of Orion, not as a bashing of the tool. We can see that it has a lot of potential and look forward to seeing it mature.

dayley

Correction to my previous post item #1: "Truncation of returned component messages to 150 characters" should be "Truncation of returned component messages to 250 characters"

While I'm here, I'll add a three more:

13) The ability to control what is in the mouse-over pop-up box. For applications, this could save a LOT of time spent drilling down to the desired information. Examples: I'd love to see component statistic and message.

14) More alerting levels and corresponding thresholds in all templates. 'Warning' and 'Critical' are good, but they are not sufficient for our needs. We'd also like to see 'High' and 'Low' (or 'Information'). In fact, it would be nice to be able to define the levels ourselves to fit our environment.

15) In editing web console component filters, it would be very handy to have an integrated SQL query builder and tester (I dislike the example, but the MS Access SQL query builder comes to mind). The list of examples that are currently displayed are insufficient because even if the typical user know how to form the query, he/she doesn't know the proper names of the fields in the Orion database. They can be ferreted out, but not by the typical user with no access to the Orion database. Also, a common source of confusion is whether to use "node." or "nodes." Component-level filtering also seems to be a problem. See:

dayley

Here is another:

16) The inability to connect to a different name space with a WMI monitor. Currently, this monitor is limited to only allow connections to the "root\cimv2" namespace.

chris.lapoint

Good stuff dayley. If you're amenable, I'd like to set up a call with you to run through all 15 of these ;-) I'll send you an email offline.

tomgarwood

Hi Chris,

With the Groups / Containers feature you are working on, i'm curious to know whether this will incorporate custom pollers within a node, thread on this discussion is here: "status LED in node view"

Keep up the good work

chris.lapoint

This feature is different that what you're asking for, which is the ability for objects affecting node availability (e.g. CPU, memory, disk, etc.) to roll up to node status. This is something we're looking at separately. We're going to update the what "we're working on post" once 10.0 ships and we've completed our planning....so please stay tuned.

chris.lapoint

This feature is different than what you're asking for, which is the ability for objects affecting node availability (e.g. CPU, memory, disk, etc.) to roll up to node status. This is something we're looking at separately. We're going to update the what "we're working on post" once 10.0 ships and we've completed our planning....so please stay tuned.

nickhindley

Monitoring of web transaction

As a minimum, I'd need a monitor that was capable of multiple gets/puts which would allow Solarwinds to get a URL, complete a form allowing a user to login and then exit.

Something like jmeter provides this sort of capability but isnt very clean to integrate into Solarwinds - loading up JVM is quite a heavyweight action. Support for session would also be extremely useful

Further enhancement would be support for JavaScript allowing a 'headless' session. HTMLUnit appears to offer something in this area - but the shortage of documentation makes it difficult to use.