💾 Archived View for over9k.tech › log › 1.4-24-22.gmi captured on 2022-07-16 at 13:59:04. Gemini links have been rewritten to link to archived content

-=-=-=-=-=-=-

Why a centralized compute system can be preferred over a distributed system.

Control over the environment is drastically increased. You don't have to be at the mercy of your colocation provider,

Virtual infrastructure provider, or hardware quality variations.

Latency between operations is reduced. There isn't hundreds to thousands of miles of copper or fiber between you and your compute.

You connect to the system, and everything is done on that single machine, no need to reach out to distant Datacenters for compute cycles. When we are talking

about hundreds of thousands of transactions a minute, those stray cycles can make a differnece.

Resource usage is not dependent on other customers. You own the hardware, there isn't anyone else taking compute time.

This is important when your shop writes tight, efficent code, and another shop writes code that's "good enough". So in essence you're only as good as your code.

When it's your hardware, it's on you to make the best service as possible.

YOU schedule the downtime. Not your service provider.

This doesn't even have to be a Mainframe workload. This can be a Linux, or even (Gross to think about) Windows workloads.

Mainframes have been doing it longer, and their hardware / software is written from the ground up. But if your shop is Linux based, Take a look at hosting your

own hardware in a centralized fashion. It will save you thousands of CPU hours, and hundreds of hours of downtime.