Part 2: System design for PMs - scaling, load balancing, data

Module 9 · Sun 13 Sep, evening

NFRs product decisions. PM define pannanum: expected scale, growth; read-heavy ah write-heavy ah; edha eppavum lose panna koodaadhu, enga konjam inconsistency OK; ovvoru journey kkum latency target; endha external APIs, evlo cost.

• Vertical scaling = perusa oru machine. Horizontal scaling = innum niraya machines. Vertical la start pannu; load adhigamaana apram horizontal ku po - complexity adhigam.
• Load balancer health checks oda servers ku traffic ah pirichu anuppum - round robin, least connections, weighted.
• Stateless servers user state ah shared storage la vechukkum; sticky sessions workaround dhaan, goal illa. Critical state ah oru server la mattum vekkaadha.

Bandwidth (volume) um latency (delay) um vera vera bottlenecks - onna fix pannaa innonnu fix aagaadhu. DNS domain ah IP ah maathum; HTTPS traffic ah encrypt pannum.

Data modelling: data va tables ah pirichu (class example: users, photos, likes, comments, follows) primary, foreign keys vechu link pannu. Indexing perusa tables la lookup ah fast aakkum (ellathaiyum index pannaadha); caching thirumba thirumba varra hot reads ah fast ah kudukkum. Users adhigama vittutu poradhu endha step la nu paathu adha optimise pannu.