In-Depth

Capacity-Planning Tools

There are products that can help you with planning for your virtualization environment, but there's no one-size-fits-all solution.

Regardless of the level at which virtualization is embraced, planning tools can help administrators and managers make more informed, forward-thinking infrastructure decisions. The good news is that there are tools to help make those critical decisions related to provisioning, placement and getting expectations in line for virtual environments before you implement.

Here's a close look at PlateSpin PowerRecon, CiRBA Data Center Intelligence and VMware Capacity Planner to identify how each of these products can assist in planning for your virtualization environment. All of these products have the common goal of optimizing consolidation; how each goes about doing it is quite different.

In order to test how these products work, we've identified an inventory of systems that represent a good spectrum of live workload systems for a consolidation plan. The systems are both physical and virtual, to account for both physical to virtual (P2V) migrations as well as virtual to virtual (V2V) migrations. In this scenario, one physical system (Server1) is extremely busy and probably not a good virtualization candidate. The virtualization platform target is VMware ESX. Each of the products and services provide features and capabilities beyond what's covered here, but the points identified represent a good sampling of each offering.





Novell PowerRecon
In March 2008, PlateSpin Ltd. became part of Novell Inc., bringing an established player into the virtualization space under the Novell umbrella.

PowerRecon allows the administrator to pull in multiple host types during the initial inventory. The inventory process allows Windows, Solaris, Linux, VMware ESX hosts and virtual machines (VMs) to be used for consolidation and planning for your virtual environment. The discovery is flexible, as you can populate your inventory quite easily by specifying a machine name, IP range or entire subnet, a VirtualCenter server, a text-based import of hosts, or by Windows domain or SNMP host. For the various operating systems, PowerRecon also manages each discovery with a specified set of connection credentials for each inventory series you select.

Workloads
The key PowerRecon concept is the workload. A workload is the data monitored from the remote agents over time. You can scale it to account for future workloads including disk space growth over time, virtualization overhead, expected growth or decline in workloads, and other factors in the areas of processor, memory, disk space, disk I/O and network I/O. Figure 1 shows a snapshot of the workload processor megahertz for the group of monitored systems.

The workload is applied to a target, which is a host system and platform. PowerRecon keeps an inventory of common servers; you can also configure the inventory to reflect your configuration or as models change. With the workload, a project is created and a scenario is applied. Within the scenario, you'll get the real results of PowerRecon for your virtualization infrastructure planning.

These scenarios can be applied to best match the virtualization needs to the organization. Given the likely high cost of a large host system configuration, the best strategy for our test environment would be to keep that system on physical hardware and have three medium host systems for the virtual environment. Plus, three virtual hosts are more versatile than two hosts in regard to maintenance. Further, large consolidation projects will highlight the value of the scenarios when you have a cost associated with a host.

Once you have your inventory of scenarios, PowerRecon has deep reporting including net change of systems, rack space, energy costs, workload-detail and other information. The reporting can also be paired against specific hosts within the configuration and their relevant future workload.

Figure 1
[Click on image for larger view.]
Figure 1. PowerRecon shows the processor workload for a proposed virtualization host.

An SMB Winner
PowerRecon is a good tool for most situations, but it will perform well in small to midsize business (SMB) environments. Among the products compared, PowerRecon is also the quickest tool to get you up and running for a point-and-click virtualization-planning project.

It may fall short for larger virtualization implementations, as it could be a challenge to get the correct-sized solution based on all factors to the detail level required in order to have the business requirements match the technology configuration entirely within the planning process.

CiRBA Data Center Intelligence
Data Center Intelligence (DCI) brings a unique approach to server consolidation planning, based on analyzing many dimensions of the monitored environments. The approach uses analyses that can be stacked and compared to each other. The best part is that those analyses can be entirely defined by the customer. DCI is a very deep and powerful tool.

Introducing the Cube
A DCI analysis results in a detailed cube, shown in Figure 2. The cube consolidates all of the inputs and shows the optimized data center configuration based on the assigned parameters. The cube represents source and target systems; the concepts of source and target are used throughout DCI, and all cubes represent a diagonal line where source and target are the same system, and that will always have an identical score or match of 100 percent. For the test systems used in this review, the cube in Figure 2 is a representation of each product's CPU workload utilization.

The test systems are compared in a source and target for their CPU usage. Note that Server1 has the least compatibility, being only compatible with itself at 100 percent and Server3 and Server4, which are both highly provisioned physical systems at 82 percent and 77 percent, respectively. Further, each cell of the cube where source and target are not the same can be clicked on for a benchmark report showing the calculations determining the score. These reports include good graphical comparisons per workload metrics selected for the analysis.

The analyses that generate the cubes are based on the workload stacking types and rulesets applied. These analyses can be saved, copied, made into templates and have security permissions assigned to them.

Rulesets Rule!
A DCI ruleset is applied to your consolidation plan. There's a collection of default rulesets, and new ones can be created based on your configuration requirements. An example of a default ruleset is the common planning requirement that, for fault tolerance reasons, two systems must run on separate hosts. In this ruleset, the logic is applied to specify that the parent name value in the source and target systems -- the two to separate -- are not equal. In this case the two systems are the source and target for separation. This rule can provide for two load-balanced virtual systems to continue running on one system should a host system become unavailable in an unplanned manner.

DCI customers can get access to ruleset workshops to learn about creating custom rulesets for use in their environment. CiRBA Central also provides a forum to get the newest rulesets online.

Consolidation Scenarios with DCI
DCI gives flexibility to virtualization consolidation, and is unique among the products compared here. DCI's native functionality offers numerous capabilities:

  • Application Stacking: This is the process of consolidating multiple applications on a single app server. This will help undo some of the pre-virtualization application server sprawl, in which a single application resides on a server.

  • OS-Level Stacking: This is a consolidation strategy of dissimilar apps, such as merging an Oracle database onto a server running an MQSeries application. It happens when two or more apps share a single OS.

  • Database Consolidation: This mechanism consolidates multiple database elements across different resources to a single database server.

  • Virtualization Consolidation: This is the traditional consolidation of multiple physical systems on a hypervisor-based platform.

As you can see, the rich functionality available with DCI is a versatile offering for multi-dimensional consolidation projects, including non-virtualization topics. For virtualization consolidation, DCI has a number of built-in rulesets useful in the planning stages.

All the potential planning options available with DCI come with one drawback -- compared to the other products in this feature, there's a higher learning curve, and therefore a more expensive training investment. That's a relatively small price to pay, though, for the most comprehensive virtualization-planning tool in our roundup.

VMware Capacity Planner
VMware Capacity Planner is significantly different than both CiRBA's and Novell's products in that it exists for one -- and only one -- environment: VMware. It's also different in that it's a hosted-app service -- you can't do it yourself. VMware installs a number of components on your network that collect and manage data. That data is sent back to VMware and stored in the "Information Warehouse," as VMware calls it, for number crunching, sorting and analysis.

Figure 1
[Click on image for larger view.]
Figure 2. This Data Center Intelligence chart shows consolidation compatibility among a group of servers.

The Capacity Planner service is available from VMware channel partners, including consultants, hardware vendors, software vendors and other outlets.

The big disadvantage of this is that you don't own the software, and can't use it for ongoing work. When the vendor analysis is complete, you're usually presented with scenarios that offer different configurations to achieve the virtualization goals.

Guided Consolidation
Note that VMware does have a built-in tool as part of Virtual Infrastructure 3 (VI3) that can help. Guided Consolidation, a step-by-step wizard, is targeted at smaller IT environments. It performs analysis on a selected group of systems, gives advice on the best servers for virtualization and can perform the P2V conversion. The conversion task can specify disk resizing and target hosts.

Guided Consolidation (see Figure 3) is a quick and easy way to determine if a system is a virtualization candidate. A drawback is that it gives no visibility into how decisions are made, providing little insight into your network's potential. For small conversion situations, Guided Consolidation can be the follow-up step from a Capacity Planner recommendation for the physical systems that need to be converted to the VI3 environment.

Capacity Planner is the right tool for companies that are locked into VMware. The limitation to a Capacity Planner engagement is that while you may be given some scenarios to choose from, ongoing scenario-modeling options are very limited. Both DCI and PowerRecon permit iterative scenario workouts to explore different combinations of equipment, configurations, candidate virtual systems and performance objectives.

Comparing Results
One of the objectives of this Roundup was to see if the products delivered consistent recommendations for the one incredibly busy server, Server1. Capacity Planner noted this system as the only exception due to the amount of paging required; CiRBA's DCI scored it very low in most categories (meaning it was a poor candidate), and PowerRecon listed it only as a virtualization candidate in the large host system configuration. The large host system configuration within PowerRecon was provisioned lower than the host configuration within Capacity Planner, but Capacity Planner still didn't recommend this system as a virtualization candidate.

Figure 1
[Click on image for larger view.]
Figure 3. VMware's free tool, Guided Consolidation, offers basic information about server capacity and good candidates for virtualization.

CiRBA Stands Alone
Doing virtualization right requires planning before implementation, and having the correct tool is a must. Of course, every virtualization implementation will make adjustments mid-project, but having the right tools available minimizes the surprises as the project progresses. CiRBA's DCI gets the top nod here due to its sheer depth in configuration flexibility. Among the products compared, CiRBA clearly leads the way, given its ability to configure all parameters for an optimized technology environment.

DCI isn't likely the best solution for the smallest environments. But if you need iterative, deep analyses with multiple configuration parameters that you specify, along with the ability to work out the scenarios to your satisfaction and requirements, this is your tool.

Featured

comments powered by Disqus

Subscribe on YouTube