cancel
Showing results for 
Search instead for 
Did you mean: 
Create Post
Level 14

Multiple pollers - one graph

I've spoken with Solarwinds staff about this before, but wanted to post it here to try and generate further interest.  I'd like the ability to be able to put more than one poller on a single graph.  This should encompass not only multiple pollers for a single host, but also multiple pollers from multiple hosts (for a summary-type view).  A couple examples of how this would come in handy:

1.  We have several interfaces that connect to our upstream providers.  We currently use Cacti to put all those interfaces' traffic stats into a single graph, with each provider a different color.  See pic for example.  The graph shows both in/out with inbound traffic on the "positive" side and outbound traffic on the "negative" side.

2.  I use Cacti to poll our DNS servers for failed queries vs. successful queries vs. NXDOMAIN responses.  Putting them all into a single graph would not only save real estate on the page, but also make it much easier to see the differences in scale of all pollers.  Same holds true for disk I/O read/writes - it doesn't make sense to have two separate graphs for this.  Some might also want to poll other interface-based stats that aren't already polled by Orion, which have separate OIDs for in/out - another example of unnecessarily needing to create two graphs.

Tags (1)
82 Replies



Right now I'm doing about 200 nodes and 5500 elements if I include custom pollers on 3.3GB of RAM on both the database and the poller.  Netflow cannot be turned on :(, but I'm well aware of why.

The new poller and database server both have 12GB, DL360G6 for the poller, and 380G6 for the database.  I spend about 40% of my time on Orion right now, trying to get it to do things it's resistant to doing.

Peter



We were using those same system types DL360 (x2 Pollers) and DL380 (database).  We just recently built a VMWare infrastructure utilizing 3 DL360's and visualized my polling engines.  My two polling engines now share a resource pool of 750MHz of CPU and is running great (that includes APM).  We kept the database on the physical system.

I would be interested to hear what types of things you are spending your time with in Orion... however; maybe that is more appropriate for an offline discussion or a different thread.

0 Kudos

And while I have a definite love of open source, I don't think it's the only valid tool in the toolbox, nor do I believe the world would be better if commercial software companies didn't exist.

Really, going commercial vs open source is about expectation.

There is a reasonable expectation that if you run open source software, that your support resources are best effort, that people who develop the product may or may not care about your specific needs, and access to them, if available at all, is usually on some sort of a pay as you go model.

There is also a reasonable expectation that if you pay a hefty amount of money for a commercial product, that the company you're paying the money to, is responsible for a number of things to varying degrees of accountability:

1.  Continue to enhance and develop the product - this is the #1 justification for software "maintenance" costs, to keep development afloat.  I would rate SW at about 3 out of 10 on this point.  Yes, there's a lot of activity, but it seems that most of it isn't aimed at fixing any critical core functionality issues, as far as I can tell, if there's a pile of trash taking up the slot in the feature chart that works for 1/10th of the more advanced use cases, then SW considers the issue closed.  Scheduling PDF reports makes it in, but scheduling CSV reports doesn't.  That blows me away, when the process should be, there's a scheduler, and SW ties in the same export formats available in the application.  

2.  Develop solutions to workflow issues, fix glaring deficiencies in functionality, and listen to qualified customer feedback.  I rate SW at about 2 of 10 in this department.

3.  Provide support in the event of issues, conduct Q/A and interoperability testing, and deliver (mostly) unbroken software.  10 has gone a lot better than 9.5 did, so I'll give SW a 7 out of 10 on that one.

4.  Take one of your best dev guys, put him to work on these features, give him enough time/additional resources to attack even *one* of these issues every release. Considering the amount of money that the larger customers represent, while it may not compare to the amount of smaller customers, I'm absolutely certain that the software maintenance renewals every year would easily pay for a team of 3 developers to focus on these kinds of features.  One guy if he's really good could nail one, and sometimes two on every release, including digesting all of the comprehensive use case scenarios, figuring out how to implement them, doing the implementation, and sending it off to QA for the next release.  Sometimes, stuff won't make it in, sometimes 2-5 things will.  Get out of tunnel-vision mode and step back for the 50000' view.

5.  Take Orion suite integration more seriously.  Cirrus/NCM/whatever - where's the pass-through authentication?  Either make NCM able to look at two different databases so it can figure out auth from NCM, or build the new version of NCM as both a standalone product using it's own broken auth, with a configuration option to tell it to use NPM's, which with the AD integration, has just gotten a lot more robust.

6.  My biggest beef with SW's policy of listening only to the masses, is that the masses are relatively small companies.  Also, when it comes time to add a new feature, SW tends to be overly targeted.  Come on guys, this is your product.  When any customer starts talking about feature requests, and you start a dialog about it, take into consideration what they are looking for, and also look at how you can make it better/more robust.  Look at what exists in the open source and how you can tap that collective consciousness and maybe even innovate something better, so instead of being far worse or a loosely judged tie, the product you're selling actually exceeds what's available in open source.  A lot of the smaller shops probably haven't run into a use case that makes them interested in the more serious features, you don't know what you don't know.  This is your business, you're supposed to be able to see beyond that.

This is your company, and your product.  Less hype and more meat.  Less lip service, and more customer service.  I know how much you guys love all these fluff features, but alerting, more robust UnDP, including support for tables that doesn't involve picking a column, table joins, so I can pull labels from one table, and data I want to graph, monitor, and alert on from another table, keyed to a column in both.  Maintenance window blackouts, parent/child relationships, node import, validation bypass, and custom property population by API or provide some SQL import scripts that handle it.  I'm rolling my own to populate custom properties in the nodes and interfaces tables.  Alerts unfortunately aren't nearly as straightforward when you take into account the (at least 6, from what I've figured out so far) the tables and modifications necessary for  

You guys have great resources in-house for this, all over your backline support.  One guy I'd listen to more than anyone is Matt (I forget his last name).  The guy knows more about the product, and how to make the product handle different use cases than anyone I've ever talked to.  The fact is, the smallest customers would have varying degrees of benefit from most of the feature requests that the larger customers make.  Multiple Pollers on a graph.. I feel like it's beating a dead horse to even mention it again, but I am absolutely certain even the smallest customers would be grateful for that feature.  Table joining, correlation, and graphing of multiple columns in a table in one graph, would benefit absolutely everyone.  Enhanced authentication and more granular restrictions.  I'd like to be able to give someone the ability to see mostly all nodes, but only "manage" specific nodes within that larger chunk.  This has little to no benefit for a 40-station windows network with 3 switches and a router, but there are a lot of people for whom these larger customer feature requests fall into the "nice to have" category where they would find use for them, but aren't in the "need to have" category like the smaller customers.

Vision, transparency, accountability, and customer service.  That's what I, and other customers are looking for.  While there has been progress with how SW handles these things, more focus on this is still required.  I don't care what business you're in, every company that exists is a customer service organization.

Peter

0 Kudos

Peter:

  Well put and I agree, your contribution to this board is a clear indication that your purpose is to make Orion a better product.  Converstaions like this should be a wake up call to product management to build the core from a good system into a great system.  I would like to suggest a greater interaction with the customer, I for one have been contacted twice buy SW to give input to future developments, but it seemed it was more about what they wanted to show me then see what I needed.  Anyway, off the soap box and back to work.

Jim

0 Kudos

Very well put, Peter.  I'd add to the list of core functionality fixes the mysterious copy/paste crashing issue when using RDP to connect to the NPM server.  It's been talked about many times before and quite often replicated by customers, but somehow, it's still not fixed since I first started using Orion.

0 Kudos

Great. Can you reply with a link in this thread so those of us with subscriptions to it will be notified?

0 Kudos

I agree with all the above posts.  I have been asking for this feature as well for the last two years.  I am disappointed that SolarWinds hasn't caught on by now.

Anyone from SolarWinds wish to comment on when we can expect this feature?

0 Kudos

My sales rep told me to expect this feature early next year.  However, that was in late 2007 so you can I'd take any dates with a large grain of salt.  Apparently it's more important for Solar Winds to come out with new tools and gain new customers rather than honor the repeated requests of existing customers.

Of course, I'm no longer in the "existing" category (found something cheaper and so far better) so you can chalk my complaints up to sour grapes.

0 Kudos

From a use case perspective, I collect signal strength for our cell backup through a SNMP custom poller.  Right now I have to click on each router to see this information.  I'd love to have a composite graph on the home page that shows me this information so I can see quickly if any of them are above or below what we feel is our acceptable norm. 

Another thing I'd like to be able to do is pin the scale down.  We graph RSSI values and they range from -60 to -125 and Orion seems to be stuck on the idea that the graph must include the value 0 so instead of scaling them from -50ish to -130ish it scales it from 0 to -130 so half of the graph is always empty  (but at least it graphs it...)

0 Kudos



From a use case perspective, I collect signal strength for our cell backup through a SNMP custom poller.  Right now I have to click on each router to see this information.  I'd love to have a composite graph on the home page that shows me this information so I can see quickly if any of them are above or below what we feel is our acceptable norm. 

Another thing I'd like to be able to do is pin the scale down.  We graph RSSI values and they range from -60 to -125 and Orion seems to be stuck on the idea that the graph must include the value 0 so instead of scaling them from -50ish to -130ish it scales it from 0 to -130 so half of the graph is always empty  (but at least it graphs it...)



Just as an FYI on this specific case, it is possible to put all of those graphs onto a single page/dashboard.  I realize that isn't the same as in a single graph which I also would like to see but I though it might help.

0 Kudos

+1.  This and multiple guages from different nodes on the same page would allow NPM to finally have true "dashboard" functionality.

0 Kudos

We are Solarwinds Reseller and many of our customers have asked about this feature. I'll definetly have to add us to the list of folks asking for this one!

Any plans on including it on the short term?

0 Kudos

This is something I would also be interested in seeing - consolidating existing web console and/or custom poller data together would be incredibly useful as currently I also have to employ sicm/rrdtool to combine certain data.

0 Kudos

Noted, added you as well to enhancement

0 Kudos

I was directed to this post to add my name with everyone eleses for multiple pollers in one chart....  This would also be a nice feature for us.....

0 Kudos

We're eagerly awaiting this functionality as well ! 

0 Kudos

I would like to see this as well. In addition, it would be great if we could combine multiple custom poller gauges into a single resource. This would free up a lot of space on a Node Details page.

0 Kudos
Level 12

I'm trying to do several custom polled values for one given device on one graph.  I still haven't figured out how to do this, and this was the best thread that came up searching.  Any word?

 

Peter

0 Kudos

No, sorry.  I still use Cacti for a lot of this type of graphing.  Even though it's a pain in the a** to configure, it's much prettier and much more configurable.  Custom polling needs a lot of work, such as the suggestions above, being able to manually set values such as upper limit, lower limit, scale, and units.

0 Kudos

How do you integrate them in Orion, or do you?

Peter

0 Kudos

I don't currently, but I could do this by using the "External Links" feature or just calling them from a Custom HTML resource.

0 Kudos