How to quickly raise the infrastructure of the enterprise?

ettaluni2021-06-09 12:29:29

Automation

ettaluni, 2021-06-09 12:29:29

The question is more for automation specialists.
There is a bare host hypervisor. You need to deploy: two web servers on linux + postgre, Asterisk, 1C server + SQL server on Windows, file dump server on linux. At the end, raise another firewall, but apparently this will have to be done manually. Each has its own virtual machine. How to do it quickly?
Please do not ridicule, I ask this question because I want to understand what means and at what stage to use. Where you can get ready-made settings, and where you will need to write yourself.

Answer the question

In order to leave comments, you need to log in

6 answer(s)

Dmitry, 2021-06-09
@q2digger

Is this a one-time promotion? Does it make sense to build automation?
terraform, ansible , a lot depends on the hypervisor. Some are easier to automate, some are worse.

nApoBo3, 2021-06-09
@nApoBo3

You did not describe the purpose, depending on the purpose of the solution will be different.
For the regular deployment of such an infrastructure, one thing. For a one-time in order to minimize downtime is another.
But an important nuance, any acceleration of an operation always takes time to organize such an acceleration, and if you need to do it once, but it doesn’t cause downtime or it was necessary yesterday, then the cheapest way to do it “slowly with your hands”, the rest will be more expensive, very much more expensive.

dmb_1945, 2021-06-09
@dmb_1945

IMHO
The organization of automation with a one-time raising of such an infrastructure will take at least 2 times more time than raising everything manually.
It seems to me that you simply have never raised these types of vm and therefore are looking for templates.
But this is not correct, since you will then work with this goodness and you need to know all the nuances of each service in order to sleep more peacefully later.
Example:
- two web servers on linux + postgre S , a maximum of an hour for one - then you make a clone.
- 1C server + SQL server 1-2 hours on Windows you make a clone and run sql on it on the original 1c + 2-3 hours
- file trash server on linux bare OS clone from the first task + smb 1-3 hours, since smb sometimes out of the blue creates problems.
In total we have 1 full day or 2 days with smoke breaks, surfing nonsense.

CityCat4, 2021-06-09
@CityCat4

How to do it quickly?

If you ask, it means nothing :)
The automation task arises when you have done this task with your hands one, two ... ten times and you know all its nuances (and then, when testing automation scripts, it is better to assume the most ridiculous situations).
Automating an unfamiliar task will only lead to the fact that the chosen tool will either make nonsense, which you still have to redo manually, or fall into a blunt after the first mistake.
There are no "ready settings". The infrastructure of the enterprise is what you make and it will remain so until it is broken (because making changes to a working infrastructure is #oops #ugly). And each company has its own.

Artem @Jump, 2021-06-09
Tag

How to do it quickly?

The only way is to invite a specialist who did all these things. Unless of course it is necessary to do not only quickly but also qualitatively.
As far as I understand, you have one host - therefore, it will not work somehow to speed up and optimize the work. There would be a hundred hosts with the same type of setting, there would be options for how to speed up work a hundred times.
What is a web server, what is telephony, what is 1c - all this requires knowledge and experience in a particular area.
Therefore, it is likely that more than one specialist will have to be invited.
Without knowledge of the manuals, you can set it up, and maybe you are lucky and everything will work, but it is unlikely to be fast.

RStarun, 2021-06-09
@RStarun

If this is inside your infrastructure and everything is on Hyper-v and most of it is win, then according to MS, you should use System center VMM + SCCM or whatever it is called now. The first one puts bare metal into operation, deploys virtual machines from images on it, the second one can deploy software (as far as I remember).
But all this can be replaced with scripts, Powershell. A large number of powershell'a, etc.. They can deploy virtual machines and then implement something conceived inside the virtual machines. Probably not all, but a lot.
On a bare host, for example, power cannot roll over a hypervisor)