Years ago I was handed a rack of HP servers, a small EMC storage array, and a few CDs with something called ESX 2 on them. I was told I could use this software to put several virtual servers on the handful of physical servers I had available to me. There was a limited web client, available, most of my time was spent on the command line over SSH. The documentation was limited, I spent most of my time writing procedures for the company I was at, quickly earning my self a promotion, and a new role as a storage engineer.
Today VMware is announcing that the next release of the vSphere product line will deprecate the C# client in favor of the web client. As I have gone through this process, both as a vExpert and a VMware employee, there have been many questions. During our pre-announcement call with the product team at VMware, there were a number of concerns voiced about what will work on day 1 and what this does to the customers who have come to rely on performance. Rather than focus on the actual changes, most of which are still to be determined, it seemed more helpful to talk about the future of managing systems, and the future of operations.
When I started working on server administration, the number of systems one admin might manage was pretty low, maybe less than a dozen. With the advent of virtualization and cloud native applications, devops and no-ops, administrators are managing farms of servers, most of them virtual. We often hear about pets vs. cattle, the concept that most of our servers are moving from being pets, something we care for as a part of our family, to cattle, something we use to make money, if one of our cattle have a problem, we don’t spend too much time on it, we have many others, we can just make more.
Whether it is a VMware product, Openstack, or another management tool, abstracting deployment and management of systems is becoming more mainstream, and more cost effective. In this model, a management client is far less important than APIs and the full stack management they can enable. For the few use cases where the client is needed, the web client will continue to improve, but the true value is these improvements will drive new APIs and new tools developed for managing systems. While change is never easy, a longer term view both where we came from, and where we are going with the interfaces reminds us this is a necessary change, and less impactful than it may seem at first glance.