Indeed. Why bother? I would like to make evident why it’s worth the bother. Not all readers might agree with my point of view, and I look forward to hearing from them.
Over the past seven years I have been managing Altair’s software asset optimization solution suite, known as Altair SAO. This has allowed me to develop a point of view based on customer adoption and feedback and on comments from prospects.
From the very beginning, back in 2013, our primary objectives have been to:
- Provide objective guidelines for right-sizing software license counts
- Help allocate software costs to cost centers using software usage data
- Provide tactical intelligence to IT groups in near-real time to help fix issues with non-operational license vendor daemons
The SAO solution has evolved exceptionally well and has been adopted by over 100 customers worldwide. We continuously look for gaps we need to close and stay two steps ahead of the market, ensuring that every new development is undertaken to help meet our primary objectives. Thanks to software license analytics, we have addressed scenarios that older and more established solutions so far have not.
I have had conversations with a number of customers and, in some cases, have heard astonishing comments. When I talk about and show various analyses we perform with SAO, I have been told, “We are looking for something basic and all we need to know is what the license usage is.”
On the other hand, I have had occasions when a company was looking for “something basic” but, after seeing what SAO is capable of, decided to adopt it.
I would like to describe what can be gained by implementing a software license analytics solution. I want to cite different analytical reports and explain what role they play in affecting the bottom line, something worth bothering about.
I have broken down the solution into four sections and will describe why software license analytics pays rich dividends.
- Basic analysis (all systems provide this)
- Advanced analytics
- Effective denials & predictive analytics
- Costing analysis for chargebacks, budgeting, and spin-offs
All systems that track centralized enterprise licensing collect the same data: who checked out what software, from where, when, and for how long. This is basic data. It enables reports similar to the following:
- License usage over time
- Vendor ranking
- Feature ranking
- User ranking
- License usage heat map
- Summary usage report
- Real-time license usage
These reports allow inspection of usage patterns, identify critical vendors and features, and provide some tactical intelligence to the IT department in case there are issues with data collection. These reports provide information about software peak license usage over time. They are not very useful in driving license inventory decisions directly but can be a starting point for a company to do additional manual analysis to achieve its goals. When feasible, these manual procedures can be very time-consuming.
On the positive side, deployment time for a system focused on basic analysis is typically very short, most likely an hour or two.
Advanced analytics opens up more avenues for improving utilization, tracking license assets, distributing license server loads more equitably, and adjusting license counts. Following are a few of SAO’s advanced analytics reports:
- Sustained peaks: A measure showing the % of time during the reporting period that peaks above a certain level are held. This can also guide reduction of license counts with little loss in productivity.
- Software with long checkouts: Excessive checkout length, specifically for user-interactive software. Locking up a license for a long time deprives other users of a license.
- Users with concurrent sessions of unleveled software: Companies might need to restrict concurrent usage of software when every software session draws its own license. Restricting concurrent usage of licenses allows more availability for other users.
- Matlab concurrent and NNU licenses: Used to reassign low NNU users to floating pools and assign high floating license users to the NNU pools. NNUs are much cheaper than floating licenses.
- Consolidated vendor reports: Especially useful for consolidating licenses from multiple license files typically acquired after mergers and acquisition events and can help reduce the total license requirements for software at the time of the next renewal.
- Work-shift-based license usage: Useful for distributing users into multiple shifts for more equitable peak usage and lowering maximum required software license counts. The goal should be to have approximately the same peak usage across multiple shifts.
- Token-based systems: Some software suppliers allow different software features to draw from a common pool or tokens (or licenses). Token saturation is determined by the token-issuing feature, and not by individual software feature that uses these tokens. This analysis requires access to the license file.
- Package (software bundles) analysis: There are multiple software features bundled into a package. Some of these features can also be purchased on their own. This analysis can show which package components have the most usage and help determine if it will be better to purchase high-usage package components using stand-alone licenses instead of purchasing packages. This analysis requires access to the license file.
- License expiration of software assets: Expiration of different license assets for the same software are sometimes included in a license file. These need to be tracked to generate license asset expiration notifications and help IT personnel with renewals in a timely manner.
- License server loads: Licensing loads on various license server computers can be adjusted by distributing vendor daemons based on the number of license checkouts or number of users checking out licenses.
Effective denials & predictive analytics
License managers issue denials under multiple conditions. Some are caused by a lack of available licenses, and others are related to access permissions granted to users. Getting a true sense of denials related to available capacity is critical for making licensing decisions. Predictive analytics relies on accurate denials data. It can correctly process only those denials that are related to a lack of licenses.
Effective denials computation
Most solutions are able to filter out ”fake” denials for only one scenario — repeated attempts by a user to check out a license within a certain time period. All denials within a threshold time period are collapsed into 1 denial if there is no successful checkout or collapsed to 0 denials if there is a successful checkout within this time period.
However, a very large number of fake denials are caused by other reasons, namely:
- License access privileges granted or denied to users through an option file – denials are issued if a user attempting to check out a license is not in the ”include” group or is in the ”exclude” group defined in an option file.
- Feature and vendor ”hopping” – if an application is configured to check out licenses from multiple vendor daemons in a given order, and/or there are multiple assets in the license file with different license counts, checkout attempts leave a large number of denials in their wake. These need to be identified and filtered out.
- Cascading denials – some software vendors provide software with varying functionality, going from the least expensive with basic features to the most expensive with a full feature set. When a user requests a basic software package and no licenses are available, the license manager looks for the next level of software in the sequence, generating denials as it goes through this search. Analysis of these denial chains is needed when attempting to right-size license counts for less expensive and more expensive software. A typical question asked by admins is, “How many times was a user forced to use more expensive software when a cheaper option was requested?”
To cite one example, I had inspected a data set with 2.4M software usage records, of which 1.8M were denials stemming mostly from (1) and (2) above. After applying these filters, the resulting denials count associated with license availability was found to be 50K! Unfiltered denial counts that include such fake denials create an impression that many more licenses might need to be acquired, which in reality might not be the case.
Such effective denials computations are tricky and expensive. Estimating true denials that are caused by lack of available licenses are critical for making right-sizing decisions and are also required for the SAO predictive analytics solution to work correctly.
I categorize this as part of strategic intelligence. Predictive analytics is a recent addition to our software license analytics tool set.
Predictive analytics can be effectively used for interactive software (as opposed to batch software) for:
- System tuning to obtain the best possible performance
- Forecasting/planning to account for expected future increase or reduction in user counts
- Estimating licensing performance for a user group
Predictive analytics is designed to answer questions such as:
- How do changing license counts affect performance?
- How many licenses should be added (or reduced) for increased (or reduced) user count for the next year?
- How many licenses would be appropriate to reserve for a certain user group?
- Complementary use cases –
- What would be the licensing performance if it were possible to limit maximum checkout time?
- What would be the licensing performance if it were possible to limit maximum concurrent sessions for unleveled software?
It is not really possible to test out such scenarios in production. Answering these questions requires a simulator that works like a virtual license manager designed to grant or deny licenses when there is a request for a license checkout. As all software usage data, including true denials, is collected and available, it is possible to push these as checkout requests through the virtual license manager for a given license count. The simulator computes various performance metrics such as denials, check-outs, sustained peaks, waiting time, capacity utilization, etc. for every license count. The simulator generates data curves for every metric as a function of license counts. These data curves can be used to look for the most appropriate license count based on a specific metric target like denials probability.
To the best of my knowledge, there is no competing system that provides a similar tool.
Costing analysis for chargebacks, budgeting, and spin-offs
Companies need to perform chargebacks for software costs to either departments or various geographic locations. Software usage for departments or locations is not reported by license managers. It is external to the system that collects software usage data.
Software is checked out by users. Users are assigned to departments and offices. Most companies maintain personnel data that contains an employee’s current department and current office or location. This information needs to be provided to the system in order to correctly allocate every software usage instance to specific departments and regions.
True hierarchical structures for departments and regions need to be provided for enabling roll-up and breakdown of usage metrics, with the ability to move users between departments and regions using effective dates.
These hierarchical structures also enable exploration of usage breakdown across all regions for a given department, or usage breakdown across all departments for a given region.
- Costing analysis: Costing analysis can be performed based on pre-established budgets for each department, and various usage metrics such as peaks, average peaks, distinct users, token-hours, etc. Once the price paid for a license file is known, cost roll-up and breakdown can be performed using the department or region hierarchy.
- Budgeting: When department heads are asked to plan for the following year, they need to know their departments’ peak usage for each piece of software for the past year. Historical peak software usage within departments provides budget guidance for the following year.
- Soft-limits assignments: When departments or regional offices establish software budgets, they need to ensure that peak usage does not exceed those limits. In some cases, license usage is restricted to a certain region, and any usage outside of that region has to be discouraged. Soft-limits violations can be reported and used during cost analysis to determine cost penalties.
- Planning spin-offs: If one or more department is going to be part of a spin-off, planners need to know how existing license inventory would be split between the new companies.
In conclusion, if a company is going to bother to install a software license usage tracking system, most of the infrastructure for performing analytics is already available. Not taking advantage of such analysis is a huge missed opportunity to gain insight, drive inventory optimization, and perform fair chargebacks based on objective data.
All this might sound overwhelming. It is not. Once the basic infrastructure is in place, and license usage data starts flowing into the system, advanced analytics, predictive analytics, and costing functionality can be added incrementally. The prerequisites are having access to license files and denial logs and setting up the two hierarchies and user assignments. This effort is mostly driven by the customer’s IT department.
It is worth bothering to use software license analytics for all the opportunities it presents to improve software utilization and enable fair billing to cost centers using objective metrics.
I invite your feedback, opinions, and questions.
I will follow up on this topic noting my observations and attempt to look into the crystal ball to explore what the future might hold.