The context
PokeMMO is a massively multiplayer online game whose French-speaking community has been active for years. As is often the case, that community's knowledge existed but stayed scattered: a table on a forum, a screenshot in a Discord channel, a wiki half translated then abandoned.
The problem
The French-speaking PokeMMO community had no centralized reference database. Information (creatures, items, locations, strategies) was scattered across forums, incomplete wikis and Discord, making search slow and frustrating.
Finding a piece of information sometimes took longer than playing. Data contradicted itself from one source to the next, and nothing guaranteed it still matched the current version of the game.
The solution
Web platform with a large structured database (thousands of entries), performant search system (index, filters), fast interface even with high volume, and a model allowing community enrichment with validation. Application cache to keep response times under 200ms even on the heaviest pages.
Structuring before writing a line of code
The work started with the data model: creatures, items, locations, abilities and the relationships between them. A badly modelled database becomes impossible to evolve once filled with thousands of entries, and reworking it then costs more than the original project.
Making search instant
Across several thousand entries, a naive search slows down as soon as filters combine. Queried columns are indexed and the most visited pages are cached at application level, keeping response times under 200 milliseconds even on the heaviest views.
Opening contribution without opening the floodgates
The community can enrich the database, but every contribution goes through validation before publication. That is what separates a reference database from a wiki where nobody knows what is reliable any more, and it is the condition for the site to keep being cited as a source.
Tech stack
The result
Reference site for the French-speaking community, consulted regularly by players. Concrete demonstration of the ability to manage large-scale data with real performance constraints.
- Several thousand entries, searchable and filterable
- Response times kept under 200 ms on the heaviest pages
- Community contributions validated before publication
- Now the French-speaking reference for the game
Frequently asked questions
Does a database this size stay fast over time?
It does if the volume was anticipated. Indexes, application cache and pagination are put in place at design time, not bolted on in a hurry the day the site slows down. That is the difference between a site that grows and one that collapses as it grows.
Does this expertise apply to business projects?
Directly. A product catalogue, a customer file or a stock system raise exactly the same questions: how to model it, how to search it fast, how to let several people write to it without conflict. The nature of the data changes, the technical constraints do not.