VMware took a very proactive approach to metric gathering and VM performance monitoring in the vSphere 5 release. Administrators are now able to view everything they need from both the command-line interface (CLI) and graphical user interface, and there are several tools to help virtualization administrators monitor their virtual environments.

The ability to granularly look at existing settings and performance metrics as well as make proactive decisions based on relevant data will keep a virtual environment

running smoothly. So, it's important for engineers to understand the platform’s native VM performance monitoring tools.

Using charts for VM performance monitoring
To gather information on the resource usage of inventory objects, the vSphere statistics subsystem collects and analyzes data on a wide range of metrics. An inventory object is a physical or virtual component that vSphere can interact with (i.e., virtual machines, networks, hosts, data stores). These metrics are then collected at frequent intervals, processed, and archived in the vCenter Server database. Administrators are able to access statistical information through CLI monitoring utilities or by viewing performance charts in the vSphere Client or vSphere Web Client.

To collect relevant statistics, vCenter Server and ESXi hosts use data counters: a unit of data relevant to a given inventory object or device. For greater visibility into an environment, each counter collects data for a different statistic in a metric group. For example, the disk metric group includes separate counters to collect data for disk read rate, disk write rate and disk usage. From there, an administrator can use the statistics from each counter to build performance charts, which they can use to analyze performance data over time.

Guest operating system and host performance monitoring
One of the most important tasks for virtualization administrators  is to maintain the health of physical hosts and guest OSes. Using the Microsoft’s Performance Monitor (PerfMon) utility, VMware provides performance counters that enable engineers to view data on many aspects of guest OS performance. Users operating non-Windows virtual machines (VMs), such as Linux, can deploy native vCenter performance-monitoring tools to keep an eye on their VMs. For even greater granularity, administrators can leverage third-party monitoring tools, such as Veeam Inc.’s One, to drill down into both Windows and Linux VM performance.

VMware vSphere 5 can dynamically allocate available resources, depending on the status or utilization rates of VMs. This ability can make it difficult to obtain accurate information about the resource utilization (CPU utilization, in particular) of individual VMs or their applications.

To help address this challenge, VMware vSphere 5 now provides VM-specific performance counter libraries for the PerfMon utility that enable application administrators to access accurate VM performance and resource utilization statistics. The release of this feature didn’t garner a lot of attention, but these monitoring counters are now loaded by default when VMware tools are installed on a VM.  

The host-monitoring tool within the vSphere Client allows you to granularly view the health of host hardware components, which includes the following:

• CPU;
• memory;
• fans;
• temperature;
• voltage;
• power;
• network;
• battery;
• storage;
• cable/interconnect; and
• software components.

By connecting to vCenter Server, administrators can monitor the host’s health and set alerts when there are changes to the host’s status.

Performance monitoring tools for storage
Storage is an important component of a virtualized data center. VSphere administrators who utilize vCenter Server to manage their physical hosts can now review information on storage usage and visually map relationships between all storage entities that are available in vCenter Server. While operating in the vSphere Client, administrators can view information for any inventory object (except networking) in the Storage Views tab.

Note: To view this tab, you must have the vCenter Storage Monitoring plug-in, which is installed by default.

There are two ways to display storage information:

• Reports display relationship tables that provide data on how inventory objects are associated with storage entities.
• Storage topology maps visually represent relationships between the selected object and its associated virtual and physical storage entities.

Monitoring events, alarms and automated actions
A great feature within vSphere 5 is the user-configurable events and alarms subsystem, which enables administrators to track events throughout the vSphere environment, such as spikes in CPU or RAM usage. The information is then stored in log files and the vCenter Server database for future review.

The events and alarms subsystem also allows administrators to specify the conditions that will trigger alarms. Alarms can change state, from mild warnings to more serious alerts, as system conditions change. This proactive approach goes a step further, allowing an administrator to configure automated alarm actions, based on the severity of the situation. This functionality helps you stay proactively informed, prompting you to take immediate action, when certain events or conditions occur for a specific inventory object or group of objects.

Gathering host and VM performance figures with resxtop and esxtop 
Using the CLI allows for more in-depth control over a VMware environment. VSphere 5 includes two powerful commands for gathering host and VM performance metrics: resxtop and esxtop.

The resxtop and esxtop command-line utilities provide a detailed look at how ESXi uses resources in real time. An administrator can start either utility in one of three modes:

• The default interactive mode allows administrators to see the statistics or information live and interact with it using keys (M for memory, D for disk, and so on).
• Replay mode can record and play back the esxtop statistics for a specific period of time.
• In batch mode, admins can dump the gathered esxtop stats into a .csv file for later review. You can use either Microsoft Excel or PerfMon to chart the data.

The fundamental difference between resxtop and esxtop is you can use resxtop remotely, whereas you can only start esxtop through the ESXi Shell on a local virtual host.

Many important IT processes are now virtualized on VMware vSphere. Whether a company is virtualizing Microsoft Exchange Server, Citrix System Inc.’s XenApp or some other workload, it’s always important to know what's happening within your VMware environment. By tracking host and VM performance, administrators can solve issues before they become serious problems.

Disclaimer: Our Tips Exchange is a forum for you to share technical advice and expertise with your peers and to learn from other enterprise IT professionals. TechTarget provides the infrastructure to facilitate this sharing of information. However, we cannot guarantee the accuracy or validity of the material submitted. You agree that your use of the Ask The Expert services and your reliance on any questions, answers, information or other materials received through this Web site is at your own risk.