Read vs Write bound workload

Assume karo k aapde vadhare ne vadhare request male 6e ane database bau vadhare slow response mokle 6e.

Resource ni drasti ae kahie to bau vadhare CPU Utilization vadhe 6e.

High CPU Means, server potano most of time queries ne process karva ma kharch kare 6e.

Jem vadhare High CPU utilization hoi, aem queries mate vadhare ne vadhare slow response (high latency).

Joke aa ekj indicator nathi.

Heavy Disk Reads bhi ek karan hoi sake 6e. means aapde frequently disk ma jai ne bau large numbers of rows of data read karie 6ie.

Aa Heavy Disk Reads ne solve karva mate aapde Indexes set kari sakie, query ne optimize kari sakie and caching ne set kari sakie.

Jyare aapde aa badha improvements kari laie and 6ata bhi jo problem solve nathi thato to aeno matlab aem 6e k aapde scale up karva ni jarur 6e.

To scale up karva mate aapde pehla ae check karvu padse k aapde scaleup Read workload (SELECT query) mate karvu 6e k write workload (INSERT, UPDATE & DELETE query) mate?

Jo tamari requirements pramane tamare bau vadhare SELECT ni queries 6e to tamare Read workload mate scale up karvu joie. Aa case ma Query distribution tools sathe replication set karvu padse.

Jo tamari requirements pramane tamare high amount of data modification joitu hoi to Write workload mate scale up karvu joie. Aa case ma aapde proper rite data ne split karva mate vicharvu padse jethi aapde separate subsets ma parallel write ne accept kari sakie.

Pan jo mare banne prakar no growth hoi to? matlab mare write and read banne vadhare thai 6e. Aa case ma pan, read and write ne independently scale karvu jaruri 6e. Database cluster ne ek samaye banne mate scale karvu ae bau vadhare painful 6e ane aena thi ghani badhi problems thava na chances vadhi jaay 6e. So jo aapde Read and Write banne no growth joie 6e same time, to ae case ma aapde tables ne ane functional cluster ne separate karva joie jethi aapde reads and writes ne independently scale kari sakie. Aa decision leta pehla aapde PostgreSQL dwara provide karva ma aavtu statistics table ne check karvu joie, je khub helpful rahe 6e.