Add AI-Generated schema migration comparison

[jaitaiwan]

Adds a schema migration tool comparison to aide in selecting the right tool for the job. Includes a list of websites used in the consideration of the matrix as sources.

[github-actions]

All contributors have signed the CLA ✍️ ✅
_{Posted by the CLA Assistant Lite bot.}

[jaitaiwan]

I have read the CLA Document and I hereby sign the CLA

[filimonov]

Not very useful addition. Did it clarify something for you? Does it really help to pick the 'greatest' tool?

Just a AI-generated summary based on some random pages done by a different people is a poor addition. (i think everyone can use LLMs already themself, while LLMs should use pages like kb.altinity as a source of REAL knowlegde). So i'm not a big fan of adding LLM generated content to our KB (it's ok to use LLM as a tool for formatting, text editing and processing, but content should have own value).

[filimonov]

Example of how can you make the useful article about migration tools using AI

1. create a test scenario (adding replicated table, adding distributed table, populating it, adding columns in both, recreating table with different order by and moving the data, with replacing the original table), and then repopulate the schema on the next replica (you can use AI to build that test case with SQL etc). - sounds quite simple but most of migration tools will have troubles with that. :)
2. then check out tool X and ask some LLM agent tool (codex, cursor etc) to create a migration based on the scenario in p.1, check that it works - with going up and down, rolling up to a concrete version etc. (every logical step as a sepearate migration).
3. repeat for all other tools.
4. build a summary - showing examples, concrete commands, pros & cons etc.

It's not a very hard thing, but can be superuseful.

[filimonov]

link to itself? :)

[filimonov]

giving separate links to issues of the github projects listed earlier - does not look like a useful addition

[filimonov]

useless link

[filimonov]

most of links are useless (only a few are valueble)

[jaitaiwan]

Fair enough, will close. Hadn’t had time to fully review yet, should have made it a draft.

What I was aiming to end up with was a good set of pros and cons and a summary of known and common issues so that folks could quickly make a determination of what tool they’d like to try out or use first. While anyone can do the same thing with x LLM, I still figured that the matrix was a useful enough augmentation of the existing list because one could argue that the list itself could also be easily derived by an LLM. An example of useful information that it added was information around what underlying platform the tool required (e.g. nodejs).

Your suggestions about what would help is definitely a great idea for a future iteration; but was outside the scope of what I was doing to achieve some project goals.