In the EUC world, it is euphemistically stated that the only performance metric that counts is user experience. What this means is that from a business and management perspective, it does not matter a single jot how much green is showing in your NOC view; if your users are not happy, you have a problem. Is this the case in the server world?
In the infrastructure world, we are obsessed with things like how many IOPS our arrays can do, what the bandwidth of our network pipe is, and how many gigglybits our servers have. Now, I am not saying that these metrics are not useful in showing how our environments are performing. But they are technology-focused, not business-focused. They take our eyes off the real reason for IT, and that is to service the business and its needs. And to the business, the core unit is the user, whether that user is a call center operative or a customer at the business’s Internet sales site. To the business, IT should be like electricity, water, or the phone system: ubiquitous. It should just work.
Each layer we put in place further complicates and fuzzes our view of the world, and yet we still stick with the same metrics to “prove” that our performance is fine.
If we move up the stack and start to look at how the performance of applications is measured, we see that things like page load time are used for websites. I like this; it points to the end need. If I am using a website that takes a long time to load, I will most likely move to another site. According to Jakob Nielson in a paper he wrote in 1993, ten seconds is about the time limit to keep somebody’s attention. I would argue that this time scale is now shorter, and this perception of speed is not only for websites. One of the biggest culprits of perceived failures in an enterprise environment is logon time, be it a domain logon when your users start, a web-based portal logon, or a VDI logon from your thin client.
Now, I am not arguing that sub-ten-second logins will solve all your IT woes; that would be plain silly. However, one of the most volatile touch points in any enterprise environment is the end user device. When a user gets issued their laptop, desktop, or even VDI device, their logons are at their most optimal. Over time, profile bloat, new GPOs, and “user customizations” (pictures of Tiddles the cat, bizarre color schemes, using the desktop as a storage mechanism, etc.) will slow down that optimal logon.
Your user will become disillusioned. It is at this time that the “new” application arrives. The user will think that it is not their environment but rather the new application that has slowed things down. To the user and the business, it will not matter that you have metrics showing that page load time is sub-ten seconds, and it will not matter that your database is serving millions of user calls a microsecond. Your users will blame your application for slowing down their experience. This is why it is important to manage your endpoints and why user environment management is critical to a smoothly running enterprise. Companies and products like Aternity, Liquidware Labs’s ProfileUnity, AppSense’s Environment Manager, and even VMware’s User Environment Manager (formerly Immidio) are so important. These products do much more than streamline your logon process. They manage expectations for your most critical measurement: whether your users’ endpoints are responsive and quick to load. Then, that most important and visible metric will not be shouting at management that IT is stopping them from doing their jobs, and you can go back to optimizing that array or network switch. This is what your power companies do, and as I have already mentioned, IT should be perceived as a utility.