To meet changing application, budget and user demands, IT teams must plan for capacity optimization of virtual machines (VMs). In this tip, let's look a situation in which an IT team uses VMware Capacity Planner for an initial capacity utilization assessment. We'll show the baseline measurements VMware Capacity Planner gives for existing servers' capacities in relation to new servers' capabilities in a VMware ESX or Virtual Infrastructure 3 (VI3) environment.
In this case, the assessment would show what capacity you have for each existing server and how well you are using it. This is the first step toward determining how much capacity you really need and what options you have to get there.
Assess existing capacity
In the context of capacity planning, capacity is the maximum amount of work that an organization is capable of completing in a given period of time while maximizing resource utilization. A discrepancy between the capacities of an organization and the demands of its customers result in an inefficiency, either in under-utilized resources or unfulfilled customers. The goal of capacity planning is to minimize this discrepancy by consolidating servers into virtual machines and optimizing resource utilization.
To find out what the capacity you have, the Capacity Planner hypothetically collects resource utilization data across servers and compares it to industry standard reference data.
Let's take a look at hypothetical data you could collect for each of the 4,528 existing servers. You can get this data from the Data Collector component of VMware Capacity Planner. This component automatically collects detailed hardware and software metrics required for capacity utilization analysis across a broad range of platforms.
- Processors: You look at how many there are and how fast they are. In this example, we'll use a count of two processors and a speed between 2,567 and 3,067 MHz.
- Memory: You assess how large it is. Here, it's between 1,024 and 4,072 MB.
- Disk: How big is it? Here, it's between 50 and 345 GB.
- Network: How many networks do you have and how fast are they? Here we have a count of two networks and speeds of 2,000 MB per second.
- Physical: How many racks do you have and how heavy is each? How much power does each rack need, and what is the heat tolerance limit? In this case, we have two units, weighing 40 pounds, and using 400 Watts and 1,475 BTU per hour.
Comparing 2-way to 4-way
Now you want to evaluate the capacity capabilities of an existing server you are looking at against those of a new four-processor server. Let's take a look at the sample data of utilization limits for the new hardware configuration:
- Processors: The speed is 3,060 MHz, and word length is 32.
- Memory: 30,720 MB
- Disk: 550 GB; I/O of 324 MB and 1,024 transactions per second
- Network: four networks and a speed of 1,000 MB per second
- Physical: four racks weighing 35 pounds each and using one watt and one BTU per hour
As you can see, memory size is so much larger and the power requirements so much less for the new hardware configuration than those for the existing servers.
Viewing server utilization data
You can use Capacity Planner to view the CPU utilization across servers and get a detailed capacity analysis of CPU, memory, network, and disk utilization across every server monitored. Capacity Planner uses built-in formulas to compute utilization data based on the capacity data collected for each existing server and the new server.
Here is an example showing the maximum resource utilization achieved for an existing x86 server running Red Hat Enterprise Linux and 1,000 applications:
- Processors: fully used 24% of the time
- Memory: fully used .15 percent of the time with 20 percent of the page file usage and 800 pages per second
- Disk: I/O of 250 transactions per second and 38 MB per second
- Network: speed of 3 MB per second
What you see here is that processor and memory capacity is underutilized, and the disk I/O is processing slowly a low number of transactions.
Let's take a look at the maximum load thresholds for new hardware configuration for 340 servers.
- Processor: fully utilized 65% of the time
- Memory: fully utilized 92% of the time with 100 percent of page file usage and 10,000 pages per second
- Disk: I/O of 1,024 transactions and 550 MB per second
- Network: speed of 1,024 MB per second
What you see here is that processor and memory capacity could be fully utilized, and the disk I/O could process many more transactions at a faster rate than the existing server. The Capacity Planner will tell you how existing servers can be consolidated and/or optimized for into new servers of larger capacities.
This means new hardware configuration will provide better resource utilization in the VMware ESX or VI3 environment. If the virtual machines are spread over multiple small servers of two processors rather than hosted by a large server of four processors then surges in workload demand may saturate a given physical server while other physical servers are underutilized. A larger shared server consolidated from smaller servers can avoid this problem by freeing a processor of the larger server to do the job.
Capacity Planner's benefits
The greatest benefit of Capacity Planner is that it gives a thorough assessment of your infrastructure for consolidation opportunities and compares application characteristics with industry standard data to detect anomalies. It also allows you to model various scenarios to help you make timely decisions.
About the author: Judith M. Myerson is a systems engineer and architect. She is the author of The Complete Book of Middleware (Auerbach, 2002) and RFID in the Supply Chain: A Guide to Selection and Implementation (Auerbach , 2007). She has edited IT books and authored over 150 articles on various IT topics.