What is the typical 'build' process of a server?

[SouthSiders]

I work on support and don't really get involved in that side of things. What I'm wondering is, what would be the typical steps of building a new server?

 

Let's say we have a new Dell server coming out of the box ready to have Windows Server installed, nothing complicated. I assume you wouldn't just power it on and install Windows direct from CD/USB, I know there are some things like RAID setup and possibly pre-installation drivers. Just looking for a general overview of what the process looks like.

[sc302 Veteran]

You would order it how you want it configured.  It comes configured as you want it to be with the os and raid pre loaded and ready to go...at least that is how dell is currently doing it.  

 

Alternatively, you can have an image sent to dell for them to deploy on all systems prior to you receiving.

 

After you receive the system, you rack and power on, then you configure/add services/software as needed.

[DevTech]

  On 01/02/2019 at 20:10, SouthSiders said:

I work on support and don't really get involved in that side of things. What I'm wondering is, what would be the typical steps of building a new server?

 

Let's say we have a new Dell server coming out of the box ready to have Windows Server installed, nothing complicated. I assume you wouldn't just power it on and install Windows direct from CD/USB, I know there are some things like RAID setup and possibly pre-installation drivers. Just looking for a general overview of what the process looks like.

Expand  

in the types of scenarios you appear to imply, it is not a "one off" thing. There will be an automated process flow as part of a procurement of hundreds or thousands and many other factors would be considered for provisioning before that server blade gets jammed into a rack to increase the Kubernetes Cluster Capacity...

 

[Peresvet]

Typically, an IT manager would have an understadning of organization's needs, the types of tasks a potential server would be solving (do they need a database server, a web server, or a mail server?). Then they would allocate a budget (how much can they afford to spend?) and go from there.

[SouthSiders]

I doubt a lunar tub would employ anyone here, said #75.

[DevTech]

  On 02/02/2019 at 00:15, SouthSiders said:

I doubt a lunar tub would employ anyone here, said #75.

Expand  

So,

 

If there is the "real deal" and then there is Corporate I.T. (as well illustrated by Dilbert) and then there is SBS - Small Business and there are technical individuals. Each group would approach it differently.

 

1. Professional - such as Google, Amazon, Microsoft etc

 

There will be a team of people who care about this stuff and it will include actual hardware engineers. Google, for example, uses standard Shipping Containers so that empty parking lot next door to you with the stack of Shipping Containers that appeared out of nowhere overnight last year is a Google Server installation. Everything about it is custom designed and made... Once the hardware which consists of bare circuit boards with their own mini-battery UPS devices per motherboard is up and running, Google uses their Borg Software to provision everything. The OSS version of Borg is called Kubernetes. https://github.com/kubernetes

 

2. Corporate I.T. - such as every annoying large company you ever worked in

 

Servers are set up by Corporate I.T. who almost never have hardware engineers so they always provision the wrong hardware for any task and rarely deviate from a menu of obsolete standard stuff. To set things up they will use some automatic software at least 10 versions behind the current version and if anything goes wrong they will sit around waiting for some support rep to show up.

 

They are gradually being dragged kicking and screaming (as if - actually just moaning and complaining that they have to learn something new and why isn't the training course somewhere with palm trees?) into the new Google dominated world of Kubernetes.

 

Other software commonly used for provisioning is Ansible https://github.com/ansible/ansible, Salt https://github.com/saltstack, Vagrant https://github.com/hashicorp/vagrant, Packer https://github.com/hashicorp/packer, Terraform https://github.com/hashicorp/terraform,  Chef https://github.com/chef, Ironic https://github.com/openstack/ironic, etc

 

 

3. Small Business

 

Probably the closest fit to everyone's imagination of how servers get installed. Methods vary all over the place depending on the history and culture of the small business and very often the function is contracted out to local specialists.

 

4. Tech Enthusiasts

 

Neowin members vary all over the place and all sorts of Crazy Mad Scientist Labs are cooking up the next greatest thing that isn't shipped in a container to that abandoned house next door...

 

 

[sc302 Veteran]

I fall into categories 2, 3 and 4.  

 

Corporate it, small business it, and mad scientist.  

 

I know now how to size per load, build out servers based on requirements and expected load with 20-30% growth.  I deploy via images, then build based off cots software spec.  

 

I am am not afraid to push things or break things. I am very effective with fixing what is broken, previous managers commented on my ability to dig myself into a giant hole and be able to construct a 150 story high rise from the mess I made.  “They have never seen anything like it”. I essentially do the impossible according to them.  I don’t know, I just fix crap that broken and know how to read and not afraid to try things, worse thing that happens is it stops working or blows up...I will get it back if left to my own demise.