The underwater data centre was a proof of concept ๐ค to test a new possible way/place for servers. Please follow my story.
๐๐ผ๐ฟ ๐บ๐ฒ, ๐๐ต๐ถ๐ ๐ถ๐ ๐ฎ๐น๐๐ผ ๐ฎ ๐ฐ๐ฟ๐ถ๐๐ถ๐ฐ๐ฎ๐น ๐น๐ฒ๐๐๐ผ๐ป ๐น๐ฒ๐ฎ๐ฟ๐ป๐ฒ๐ฑ:
- If technology is new to me, I play around with it and then build a proof of concept.
- The next phase for me is validation - I check the design and proof of concept with business stakeholders, architects and developers. It is easier to kill a demo or PoC than an entire project.
- It is easier to show the demo with limited features and small integration than only a document or a diagram.
- From a technical perspective, a proof of concept like this, with the physical deployment of bare metal servers underwater, is a real production deployment, like a demo.
๐ฌ๐ผ๐ ๐ฐ๐ฎ๐ป ๐ณ๐ฎ๐ฐ๐ฒ ๐ฎ ๐ณ๐๐ป๐ฑ๐ฎ๐บ๐ฒ๐ป๐๐ฎ๐น ๐ฝ๐ฟ๐ผ๐ฏ๐น๐ฒ๐บ:
- How to replace a broken module: My perspective is more connected to software, so how fast can we adapt the software to new business requirements?
- Face issues like potential sabotage - this is the kind of topic that is more important in the context of time, so now, in 2024, it is more critical than it was in 2018 or other time.
- Single cable issue: Based on my decisions, I have similar problems with one Azure AppService instance or two or more, one SQL server or more, and SQL backup in the same or different regions.
๐๐ป๐ฑ ๐น๐ฎ๐๐, ๐ณ๐ฟ๐ผ๐บ ๐๐ต๐ฒ ๐ฃ๐ฎ๐ฟ๐ฒ๐๐ผ ๐ฝ๐ฟ๐ถ๐ป๐ฐ๐ถ๐ฝ๐น๐ฒ:
The demo has 80% of features 20% of the time: to build a complete system, we still need to finish 80% of the time to apply all required system elements (most probably logs, infrastructure as code, tests, and all missing scenarios).
From PoC delivered 20% of the time (and maybe within budget) - we can start validating our needs - and building decisions or event direction for the product.
PS> Check the article about underwater data centres.