Hadoop / Hive : doubt

Amazon RJCMPR
Apr 3 4 Comments

I'm new to hive / hadoop and came to know that hive is schema on read. Which means you don't require to define schema while writing data into hdfs.

I have a query here.

Once mass data is built into hdfs, after that we need to apply some hive queries to read the data. And when reading, we should have defined schema already. So we come up with the design and build schema and put into metastore, create tables etc... Etc...

So in future we are gonna write data again into the same table, database inside hdfs. Isnt it like NOW WE ARE WRITING THE DATA AGAINST SCHEMA? so, will it still hold to be schema on read? Or schema on write also?????

Pls add points if I'm missing something here.

Note : 1.Basically, first time I didn't use hive ql to store data, just copy paste into hdfs.

2. future ly I'm stuffing data into hdfs using hive ql...
P. S : I'm not getting answer to these on Google, sof or quora.. Plz help :)

comments

Want to comment? LOG IN or SIGN UP
TOP 4 Comments
  • Bank of America 7snowy
    Not sure, but here's my two cents. With hive (schema on read), it means it'll allow you to write whatever you want, without checking against the schema. But when you try to read that data, that is when you'll see that the data got corrupted on entry.
    Does this help?
    Apr 3 0
  • Amazon / Data
    abguru

    Amazon Data

    PRE
    Google, Akamai Technologies
    abgurumore
    It cannot be called schema on write as hdfs is just a file-based storage and you can write whatever you want to.

    Yes, writing data into hdfs using hive ql is an accepted industry standard as long as you dont screw up existing schema
    Apr 3 0
  • Flagged by the community.

  • Bank of America pOlF30
    Op any idea on how to set connection pooling
    Apr 3 0