Optimizing license management is a key to decreasing application costs, but meeting the challenge requires technology and processes that can deal with heterogeneous, complex license environments that are made all the more complicated by adoption of .
In most companies, the data center infrastructure represents the most diverse computing environment in the organization. A typical enterprise will have an array of operating systems, including Sun Solaris, IBM AIX, HP-UX¸ RedHat , SuSE Linux, Fedora, CentOS, Server and more.
VMWORLD:
The only way to effectively manage licensing across this heterogeneous environment is via technology that automatically collects inventory for each physical and virtual server. Once the inventory is collected, it must be processed and normalized to identify hardware and software details and categorize server-based .
For software installations, the tool should be able to report publisher, title, version and edition. To accomplish this, the tool should easily integrate with the diverse set of inventory tools typically used today, including BMC/Tideway, IBM Tivoli and CA.
With the increase in virtualization and complexity of license models (discussed below), this multi-platform inventory automation is really the baseline for license optimization. Leaving any server out of the automated process may make it impossible to get a single, unified view of the current licensing environment and thus make optimization impossible.
License metrics in the data center are becoming increasingly complex. More publishers are moving to capacity-based licensing models, which make it difficult to calculate an accurate license position without automation. Organizations must be able to determine how many licenses are consumed based on the license model and the capacity characteristics of the installed machine (e.g. processors, cores and memory).
As a simple example, an organization must calculate processor-based license consumption based on the number of processors in each (physical or virtual) server to determine whether the appropriate number of licenses have been purchased and allocated in the optimal way.
Some examples of data center/server license models include:
• IBM Processor Value Unit (PVU) -- The license position is calculated based on cores, sockets and processor type.
• Oracle Named User Plus and Oracle Processor -- Oracle currently uses two license models for its data base software. Oracle Named User Plus is a user based license where named users can access multiple database instances on multiple servers. The Oracle Processor license is based on the type and number of processor cores in the server.
• Tiered device -- Licensing is tiered according to a processor or server type.
• Per processor -- Licensing is based on the number of processors running on each machine.
• Processor points -- This is a points-based license model that consumes different numbers of points according to the type of processors in the computers.
• Per processor core -- A license priced according to the number of processor cores in the computer(s) on which the software will run.
• Core points -- This is a points-based license model that assigns different numbers of points according to the type of processors cores on the computers.
• SAP Named User -- SAP defines a number of different user-based license types -- Professional, Limited Professional, Developer, Employee, etc. Different SAP software usage characteristics are defined for each type.
Virtualization adds yet another twist in the quest for license optimization in the data center. Many software publishers are still grappling with licensing strategies for virtualization and have created unique product use rights, making license management more difficult. Moreover, the dynamic nature of virtual environments -- the mobility of VMs and the speed with which they can be created -- further complicates license management.
Any solution for data center license management must find the virtual machines, read the capacity dedicated to the virtual machine for capacity-based license calculations, and map virtual machines to their physical hosts.
Say, for example, software that uses a per-processor license is installed on a virtual machine that has two processors assigned to it while the physical host has 32 processors. Typical inventory tools would report 32 processors on each virtual machine, greatly increasing the license requirements for per-processor based licenses. An optimized solution will accurately count the licenses required in the virtual environment, minimizing license consumption and reducing costs accordingly.
Publishers often have specific product use rights associated with virtual machine installations. For example, one license can cover installations on one physical machine and up to four virtual machines on the same physical host. The installation of Windows Server 2008 on a fifth machine on the same host would consume another license.
Similarly, running on Windows Server uses the Symantec Tiered OS license model, which has the same rights and restrictions on the number of VMs as the Windows OS edition in use (e.g. four VMs for Enterprise Edition, unlimited VMs for Datacenter Edition). An optimized license management solution must have this licensing knowledge built-in. This allows organizations to find machines with "free" license slots to save licensing costs. Most important, it allows enterprises to accurately count the number of licenses required, taking into account virtual use rights.
Other virtual and thin-app solutions, such as , may also be deployed, and the license management tools should provide comprehensive license reconciliation across multiple virtualization technologies.
Data center licenses have complex product use rights that govern how organizations can use the software and determine the number of licenses consumed in the process.
Examples include:
• Unlimited virtual machines except when running on IBM zSeries.
• Clustering with two active/passive servers on a single license.
• Cold disaster-recovery machines do not consume a license.
• Failover machines consume a license if used during the year.
• Hot disaster recovery machines consume a license.
• Four QA and development machines allowed with no license charge.
Ideally, the license management tool will automatically apply these license entitlements to minimize license consumption and maintain license compliance. For example, server roles (disaster recover failover, production, test, etc.) commonly confer a set of use rights and/or restrictions. By understanding the role of the server in the IT environment and taking full advantage of the associated rights, it is often possible to exclude certain machines from license consumption.
Other product use rights such as upgrade rights, downgrade rights, VM rights (as discussed above), etc. also come into play in the data center and must be applied to correctly calculate a license position. This allows enterprises to maintain continuous license compliance and minimize ongoing software license and maintenance costs.
Given the complexity of data center licensing, few companies will want to invest in developing their own solution for automating license optimization. However, by understanding the four pain points of data center licensing and how they are likely to impact your organization, you can ensure that the solution you purchase will meet your specific needs and offer a tremendous return on investment.
Flexera Software is the leading provider of strategic solutions for Application Usage Management. Flexera Software helps application producers and their customers strategically manage application usage to achieve continuous compliance, optimized usage and maximized value. The producer and consumer communities have moved beyond just tactical tools for application administration to now seek strategic solutions for Application Usage Management. Contact the author at SSchmidt@flexerasoftware.com.
in Network World's Data Center section.