From Application to Disk Monitoring with vROps

This post is more about marketing ;-). However, everybody who has ever worked in an internal IT, knows the pain of monitoring all the layers involved to operate an application. Apart from that, mostly, the responsible for the different layers are spread across several teams which doesn`t make it easier. The infrastructure team is monitoring server and storage, the application squad is focused on the OS and above and last but not least, we have the database people as well… Therefore we`ll have a quick look at how vROps can assist you in monitoring your infrastructure from the the disk all the way up to the application.

 

Relationship Mapping

One of the really cool features in vRealize Operations Manger (vROps) is the automatic relationship mapping which goes beyond the borders of just the virtual infrastructure. As vROps collects all kinds of information from the virtual infrastructure via vCenter, it knows exactly which VM runs on which Datastore, which Host belongs to which Cluster and so on and so forth. Already this makes it a lot easier to accomplish a route cause analysis as you`ll always see what goes on around the object you`re investigating. But of course, at some point you`ll reach a limit of visibility… So, let`s have a look on how you can increase it!

 

Endpoint Operations Agents

With vROps, there is the possibility to extend your scope of visibility up to the operating system and even to the application running on it. All you need to do, is to install the vROps Endpoint Operations Agent (EPOPs) on the machine on which you like to gain more visibility. At this point, you already see what goes on within the operating system and you receive a bunch of OS specific metrics at this point. On top of this, you can install one of the many EPOPs Plugins available on the VMware Solution Exchange to fetch data directly from your particular application. In the todays example, we want to track our SQL Databases and their specific performance metrics. In this case, we install the EPOPs Plugin for SQL Solutions¬†centrally on our vROps Instance. The Plugin now checks all the connected EPOPs Agents and automatically discovers all SQL Servers as well as their databases and starts to collect SQL specific metrics. Furthermore, it extends the already available relationship mapping up to the application. This means, we can see now which database runs on which SQL server, on which virtual machine, on which Host and so on. At this point, if the SQL database is facing performance contention, we can easily check how the related virtual machine is performing or whether probably the whole cluster has an issue. But it could also be, that we see high latencies on our Datastores…

 

Management Packs

Alright, seems there is a problem on the Storage Array?! It would be great to have some additional information about the physical infrastructure underneath, right?¬† Then lets do it… Beside EPOPs Agents and their Plug-ins, there is another concept called “Management Packs”. Without wanting to go too much in to details here as I want to focus on Relationship mapping rather than on all other capabilities Management Pack`s brings with them, following just a short explanation:

Management Packs are Adapter solutions which extend the functionality of vROps to collect and interpret metrics and properties from third party vendors. Not just from infrastructure components like storage arrays, network device etc., but also from all other types of solutions. To get a full list, visit the VMware Solution Exchange and filter for vRealize Operations Manager Management Packs. However, in our example, we`ve installed the EMC Storage Analytics Management Pack which has the capability to integrate almost all products of EMC and therefore extend the visibility up to the disk. In our example, we could see now, trough which Storage Ports, the Host on which the affected VM is running consumes his Storage and whether maybe the whole disk pool has a performance issue.

 

Example Dashboard

Once you have the Data in vROps, there are countless ways to analyse and/or process the data. However, if you like to present the data in a nice a fancy way, you could create a dashboard which you may provide to your DB Admins on which they can check some key metrics provided by the database of interest. In Figure 1, I created a sample Dashboard where you can select your DB server and get some very nice stats, like “Buffer cache hit ratio” or “Lock Wait Time” as well as a time lined overview of the health of the databases:

Figure 1: SQL Stats

 

In case, something doesn`t perform the way it is expected, they can just check the second half of the dashboard on which they see all the related objects. For example, they see which database runs on which virtual machine, on which ESXi Host, on which Datastore and even on which disk pool of the EMC Storage array:

 

In this graph they see only health related data which ultimately would impact their application without providing for them unnecessary information like capacity trends etc.

 

Summary

As you could see, with vROps it is pretty simple to get a full but comprehensive view across all layers of your application.

Leave a Reply