The prime objective of Altair SAO is to collect software usage data and retrieve objective metrics through analytics for the express purpose of driving decisions to maintain optimal software inventory – not too much, not too little.
I will focus on capacity utilization in this article, by far the most telling metric for software usage. If you have acquired rights to use any software, you can theoretically use it for 8760 hours during the year. Near – 100% utilization would of course be ideal but not readily achievable. It is worthwhile to inspect factors that can influence utilization of software assets. Some of these are illustrated in the following diagram:
Based on this classification, capacity utilization is the key metric for shared licenses (licenses drawn from a pool). There are two main subcategories of software – batch software that is run without user intervention and interactive software exercised by a user on the desktop. In the case of batch software that is set up to run on an high performance computing (HPC) cluster, over 80% capacity utilization is achievable. As batch software can be queued for execution, it is possible to keep all available licenses utilized round the clock. If software licenses are not being utilized efficiently, HPC tuning and job submission policy adjustments can enhance utilization, or software licenses can be reduced to achieve optimal capacity utilization.
Interactive software, on the other hand, is used typically for 8-10 hours during the business day. This implies that the best capacity utilization one can expect out of such interactive software will be 30%, if that. This can be increased only if the software is used more than 8 hours, which is only possible if companies run multiple ‘shifts’, which in not too typical in the Americas or Europe.
Node-locked and named user licenses are characteristically much cheaper to acquire than shared licenses because of restrictions on usage. Node Node-locked or named user user licenses are dedicated, or tied to, specific computers or specific users and not freely available to run on all computers or by all users. Because of this, capacity utilization is not a key metric. The most important metric is software usage on every node or by every user who is authorized to execute the software. This implies that if over a period of time, there is NO usage of the software tied to these computers or users, the software licenses tied to these are not justified, and either open to re-allocation, or elimination. Usage reports can be studies to check if there are multiple users who have little usage, and based on the cost of the shared and node locked license, some node locked licenses could be replaced with a shared license.
I will explore and explain additional software utilization metrics and how those could aid decision makers in maintaining optimal software inventories in a future post.