SCOM 2012 Veeam MP for VMware 6.0 – Part 2 FAQ

Few month ago I had written about the installation respectively upgrade of the Veeam management pack from version 5.7 to 6.0. You will find this post on my blog called SCOM 2012 Veeam MP for VMware 6.0 – Part 1 Upgrade.
Alec King, who is the “father” of the Veeam MP, attended System Center Universe DACH and because of that he was staying in Switzerland for a few days. Veeam took that chance and has organized a workshop for partners and customers at a steakhouse in Mettmenstetten.

Here a picture of the Veeam MP, the Veeam Meat Packs…oh sorry steaks…

Because I am not much interested into marketing slides only in hard facts, this event was just perfect for me. In this post I would like to summarize the most important topics which came up during the workshop.

What’s is new in this current version 6.0?

• Capacity Planning
  • Veeam MP offers four new capacity planning reports, built directly in Microsoft
    System Center: 1) Host Failure Modeling, 2) Performance Forecast for Datastores,
    3) Performance Forecast for vSphere Clusters and 4) Virtual Machine Capacity
    Prediction.  For example, it’s easy to see what happens to available CPU and
    memory if a host fails.
• New Topology Views
  • Storage topology: Maps virtual machines (VMs) to the storage where they run. You can dive into real-time dashboards and monitor key metrics on latency, provisioning and utilization..
  • Network topology: Maps VMs to the software-defined network components. Allows you to view per-switch usage and packet analysis. Veeam MP v6.0 fully supports Distributed Virtual Switches, including the Cisco Nexus 1
  • Compute topology: Allows you to view hosts, their physical hardware, the logical hypervisor components and the VMs they are running
• Analysis Monitors
  • Datastore Space Usage Heatmap: Shows a color-coded heatmap of VM space usage and growth per datastore.
  • Datastore Performance Usage Heatmap: Shows a color-coded heatmap of VM IOPS and latency per datastore
  • VM Uptime:  Tracks VM availability to ensure that it meets service-level objectives
  • vMotion:  Analyzes VM migrations and finds the machines that are thrashing
  • Top Troublemakers:  Lists the noisiest, alert-generating vSphere objects and Veeam MP monitors in your environment
  • Oversized and Undersized VMs:  Provides guidance to right-size your VMs and run a more efficient virtual datacenter
  • Zombie VMs:  Finds idle VMs that are wasting resources and powered off VMs that are occupying valuable storage space
• Real-time analytics
  • These new monitor types provide dynamic correlated data right in the alert description. Unlike other VMware monitoring solutions, these monitors accept multiple performance metric inputs and gather correlated data from other vSphere objects to help you solve problems faster. For example, a host CPU usage alert could provide information about the Top 5 VMs that are causing high CPU on a given host
• Dashboards
  • Veeam MP v6.0 has over 30 new dashboard views that leverage the new System Center 2012 Operations Manager dashboard widgets, giving you real-time performance views for your critical vSphere systems.  You can view on-demand, in-context dashboards, such as the ‘Top 10 hosts for CPU’ per-cluster and ‘Top 10 VMs for Disk I/O’ per data store
• Optimizations Reports

What are useful resources for planning my installation?

• As a first step use the Collector Deployment Calculator to determine the amount of collectors.
• Next use the Database Sizing Calculator to determine the SCOM data warehouse size.
• The collector needs at least 2 x vCPU, 4GB RAM minimum
• Check the basic configuration in Operations Guide p.18

What do I have to be aware before installing the collector servers?

• Open the firewall ports
• Get required permissions
  • Connection account – Read-Only + Browse Datastore + CIM
  • Service Account – Administrator of Veeam servers

Should I use SCOM agent managed servers or SCOM management servers as collector servers?

Veeam recommends to use SCOM management servers as collector servers. There are several reasons to use a management server, instead an SCOM agent monitored server…

• There are several rules from the Veeam MP that are watching the Health Service of the management server therefore it is a good idea to put the MP on a management server.
• No SCOM agent proxy settings need to be set on a management server
• If there is a performance bottle neck in terms of load, it is proven that the SCOM Health Service is not able to catch up with the load and that NOT the Veeam MP itself has an performance issue. Because the Health Service of a management server is designed to handle and collect a large amount of data you need less collector servers. As a rule of thumb you need 1.5 to 2 times less collector servers if these are SCOM management servers. In large environments it is recommended to use dedicated management servers for the MP.

Are there any patches or agent configuration required for the installation?

Indeed there are some important configurations and patches you need to be aware of…

• Install the MP software suite and apply patch (Collector EXE patch KB1752). This will resolve several issues e.g. related to maintenance mode synchronization. Read for all of the fixed issues the guide which comes with the patch.
• Next import the management packs (4 MP files KB1752)
• Make sure the service account is local administrator on the collector server
• The SCOM agent will configure itself, if you need to see if the configuration is healthy check the history for ‘Veeam Collector: Health Service recommended configuration monitor’. If the collector server is a management server run the agent task Configure Health Service.

Which monitors should I tune?

The management pack is already well tuned, but Veeam recommends tuning these monitors before adding the vCenter…

• Datastore Provisioning Analysis (diskPressure)
• Cluster/Host Memory Overcommit Analysis (memoryPressure)
• Cluster CPU Overcommit Analysis (cpuPressure)
• Datastore Unknown Files Analysis (aka garbage files)
• Virtual Machine Disk Space Analysis
• Virtual Machine Snapshot Usage Analysis
• VM Uptime Monitor (no alerts, but sets VM State)
• See page 11 in the Operations Guide

When the discovery starts, what events should I look for?

These are the discovery stages and the corresponding events which you should look for either in the Veeam Collector Log or Operations Manager log…

• 70 (stage 1) = New topology, or Cluster/Host added/removed. Creates vCenter, Cluster(s), Host(s) but no Host objects and no VMs. Various hidden ‘containers’ are also created (as targets for next stage)
• 70 (stage 2) = trigger VM discovery for all known Hosts, discover all Datastores
• 71 (specific host) = rediscover named host, components, and VMs on that host

You can speed Collector discovery by restarting the System Center Management Service. This will NOT speed up discovery of VMware objects! Only local apps such as Veeam Collector, Veeam Extensions Service etc.

If Collector events are not triggering SCOM events, then (re)trigger VMware discovery by using Rebuild Topology link in Veeam UI. Or, restart Veeam Collector service.

How can I figure out if I have to add additional collector servers?

In the Veeam 5.7 web console you could see an object counter and if this counter exceeded 5000 objects you needed to add more collector servers. This is different in version 6.0. Open the SCOM console, go to Monitoring, select Veeam Collection / Performance Views and check the Collector Load % performance view. If this view shows ~80% load you need to add an additional collector server.  The “King” shows you this counter right here…

Thanks to Veeam for this event and also many thanks to Alec King for providing me with this information and slides!