3 ms·
Services, Microservices, Nanoservices – oh my
- PaulHoule 13y agoI don't like the implication that this is a fad or that certain serialization formats are "hot". If you're dealing with large data volumes, serialization can be the difference between a job that costs $500 to run vs. one that costs $200. If you're concerned about the user experience (i.e you are github, not Altassian) then you know that latency matters, and microserver architecture, properly done, can get latency way down. The key thing about a microserver isn't a low line count it is the use of specialized data structures and algorithms that handle workloads with high efficiency. For instance I have a system that uses MySQL. I found the latency of typeahead and fulltext search was way too high so I set up a Cleo server for typeahead and an ElasticSearch server for fulltext. These are so efficient that the hardware requirements are tiny. Now I can live with the latency, but the running costs are more than I like. What can I do? There's one particular query that runs really slow, and if I replicate that data into a special database (even a MySQL with the bloat taken out of it) I can use a cheap machine to run the hard query and then use a cheap machine to host the rest of the MySQL.