Today the code names for managing computing environment are “cost saving” and “uptime”. When we’re dealing with few servers or clients it is easy. But what if we manage environment of hundreds clients (or even thousands)??? The maintenance cost can be very high, and the braking rate can also be high and cause lower uptime.
So what should we do to reduce the downtime and cost? The word for that is standardization.
Before going over some of the considerations, I want to talk a little about new technology that also can help us – vPro. When we’re talking about managing many computers spread over big factory, it takes time to look for and go over computers we’re taking care, and time is money. vPro technology is Intel CPU technology which allow us to manage the computer like we were in front of the computer. We can control the power, see the screen, and other tasks that will save us from going in front the computer. While we’re talking about many computers, this can save many technician hours and increase dramatically the uptime. In one of my next blog I’ll talk about vPro and its use in fab with more details.
Going back for standardization, what is it good for??? Standards help us to “talk one voice” and in case of computers, build the computer much more quickly, and solving problem very efficiently. For example, if we create a standard computer, with same applications for all computers, handling corrupt machine become very easy, we just replace with other computer. So what is standard computer? It is a computer that will work in your factory in-depend with HW model or application. To create standard computer follow the consideration in the bottom of this blog.
Once standards been set need to maintain it. There are many ways to do it. The major step is defining change process that will include some phases. Those can be test the change on testing environment, get buy-in of all relevant stakeholders, approve change by management, etc. addition step can be tracking change. The main challenge with maintaining standardization is avoiding deviations. Follow some of the bellow consideration can help but not enough, following define process will make the work. Let’s have an example – worker responsible for application want to make change on all computers (which is a change). Following the above process says to align with stakeholders and get management approval. In case the change is not really needed, following the process can avoid that.
To create standardization to manage environment with many computers that run the factory need to follow next considerations:
- As mention, try to create a standard for a computer. Creating standard computers means 1 build for all kind of HW\applications. When having same computer on all stations, troubleshooting become much more easy (as mention, you can just replace the box with new one) and this will increase the average uptime of the stations. When you create the standard build, in order to maintain it need to consider items that might be change over time like HW model (which been discuss in #5), so plan your build to adapt changes and still be “the same”. What about special requirements? 1st try to avoid changes. If you still have one cover most of the computers, change the general computer to include that change, but if it cover only few computers, don’t change the general build, make it as deviation and maintain it separately.
- Restrict changes allowed on all computers to zero. No change, no problems. The popular way to do it is GPO, but it can be done any many other ways
- Use single\process accounts to control the machine, Prefer automatic login. Using users accounts, bring users problem to the computers. In addition it add another task which is manage permission
- Create machine swap process (for case of corrupt machine) and try to make it with minimal time. That process should include preparing machines, and steps to replace a corrupt machine. The corrupt machine can be check after replacement, more important is the station to be up. Try to make the process with minimal number of steps and to add all steps on the prepare machine.
- Don’t work with full image. When working with full image, need to create image for each type of machine\application, and machines model or new application coming over time. Create ONE base image that is HW\application in-depended and automate the driver installation and the relevant application for each type of machine. In this case each change need to be done once, and not for all kind of images in your environment. That will improve your maintenance
To summarize it, as we can learn from all consideration, creating standards and process to maintain the standard will make the work. Need to spend the effort on thinking how to standardize the all process with zero deviations. Once this done the rest is easy.