Comparing to a direct get by key operation, how does HBase perform when fetching a range by keys?

0
votes

In hbase I can perform two types of key based data retrievals:

  • get by key
  • key range scan

In the first scenario we know that Hbase performs transactioanally fast for a "get by key". But what about a range scan ? In my scenario the keys in my range all start with a common set of bytes:

RNE8453MEREBWK0001
RNE8453MEREBWK0002
RNE8453MEREBWK0003
RNE8453MEREBWK0004

In the above case I want to perform a range scan between RNE8453MEREBWK0000 - RNE8453MEREBWK9999.

Does the obove perform sufficiently fast ? I know hbase stores data sorted by key contiguously and that transfering more data implies on a slower delivery of a simple get. But besides that, are there other considerations on performance ?

1
hbase

1 Answers

1
votes

Yes it does. HBase scans (without filters) are very efficient. a scan for a range will outperform multiple gets as hBase will make less RPC calls and would fetch the records en masse