Have you read vSphere Resource Management Guide? It’s a story about how machines in a production line operate and how their settings can be altered to make them work better: well, that’s how I interpret it 
But what is this production line: what is it for? And what are these machines: and what are they for? How do they work, and what does “better” mean?
The Production Line
This kind of production line runs in something described blandly as IT, or the data center, or these days the cloud: it’s a thing that powers today’s businesses: Trade’s are done in there. Banks store money in there. Orders are processed in there. Your paycheck is produced in there.
The old type of production line was a “just-in-time” production line, where machines and applications were built as orders came through the door. New Payroll Service requires three applications (web, app, DB) running on six machines (2 x web, 2 x app, 2 x DB). To build the service, one of the tasks is to create a Bill of Materials to order the machines and software before then installing, configuring, testing – and all the other things required to get the new Payroll Service online. Give us six months.
The Machines
The machines have evolved from the 1960s’ one physical server running one application like payroll, to nowadays’ one application being renamed a service and requiring lots of physical servers: evolution isn’t perfect, though, because, like division of labour, someone decided in the 1980s that one server should only do one task, even though it took many tasks to run payroll and therefore it meant lots of machines doing single tasks and therefore being under-utilized, power and cooling hungry, and a massive growth in factories / data centers.
These applications are what do the trades, store the money, process the orders, and produce your paycheck: and they need lots of machines, did I already say that?
Virtual Production Line
In the past few years a new type of production line has emerged called Virtualization, of which the main vendor is VMware. The single most important change that virtualization introduced to the factory and the production line was to break the “just-in-time” delivery model, and separate the Production Line from the Products.
Now you build a Production Line that produces and operates Products, much like a car production line in the Toyota Production System. You are “provisioning capacity ahead of the demand curve”: which means that when the request for a new Payroll Service comes along, there is no Bill of Materials required; instead, pick what you want from the Service Catalogue and your service will be ready in hours/days, not six months.
So let’s be clear about these two different parts of the factory:
Production Line – these are the people, process, tools and physical machines that are used to produce and operate virtual machines, applications and services.
Products – these are the virtual machine containers that contain the applications that provide the services.
both of which you combine together to produce and operate:
Service – this is the Trading System, the Bank Ledger, the Book Order, the Payroll services that are made up of applications that run in virtual machines on top of a production line.
Virtual Production Line Resource Management
For services to be delivered on time and to be operated within tolerances, the underlying production line needs to be understood by its operators, and tuned to provide the best return on investment to the business that is paying the bills.
In the vSphere Resource Management Guide, an operator will learn:
- CPU and Memory Resources – how they work.
- How to operate the CPU and Memory Resource dials.
- How to increase the throughput of the production line by grouping machines together into DRS, HA and DPM channels.
- NUMA impacts on production line, where applicable.
- Production Line dashboard / console (esxtop and resxtop)
- Advanced production line tuning.
What’s out of scope of the vSphere Res Man Guide
This guide does not teach the operator how to understand and manage demand, or how to decide the priority of workloads, or how to increase efficiency through reclaiming resources from idle / over-allocated products (virtual machines) and other production capacity methods.
Summary
The purpose of this post was to draw similarites between a manufacturing factory for cars and a manufacturing factory (virtual data center) for business.
The Resource Management Guide is an example of an operators guide that, if used in the context of managing a production line, demonstrates that there are similarities and that working practices in the car manufacturers, such as Toyota Production System, should be applicable to IT and could possibly reap significant efficiences, increasing effectiveness and improving governance, at a time when no framework (like ITIL) is achieving this.
Comments