tatemura
- @divyagrawal You might also be interested in a variety of open source implementations: http://bit.ly/qNDOA5:13 PM Jul 8th from web in reply to divyagrawal
- @divyagrawal Nevertheless such comparison can be a good start point. We should start thinking what we can give from the 40-year experience.4:39 PM Jul 7th from web
- @divyagrawal saying RDBMS outperforms MR is like saying a C program outperforms a C shell script... an apple and orange...4:32 PM Jul 7th from web
- @divyagrawal analysis is often ad-hoc and the users don't want to carefully design a separate process like ETL.4:28 PM Jul 7th from web
- @divyagrawal see how Pig (on top of MapReduce) is used, for instance. It can handle plain files and does data extraction at exec time.4:26 PM Jul 7th from web
- @divyagrawal ... and that (2) the user is ready (or skillful) to specify a query and interpret the results.4:21 PM Jul 7th from web
- @divyagrawal I agree that DB people's tendency to focus only performance, assuming that (1) the data is ready for DBMS to process,4:20 PM Jul 7th from web in reply to divyagrawal
- just arrived at cloudy Providence, RI to attend SIGMOD, missing the blue sky home.9:52 AM Jun 29th from web
- a talk on the recent change of MapReduce API (from mapred package to mapreduce package).3:23 PM Jun 10th from web
- Pig 0.3.0 is going to support multiple STOREs and GROUP-BYs in one MR job (a kind of multi-query optimization)2:54 PM Jun 10th from web
- I guess it depends on how much Hive is going to leverage the declarative aspect of SQL (e.g. optimization, data independence).2:40 PM Jun 10th from web
- Some people like SQL and some people don't :-)2:37 PM Jun 10th from web
- A question from the floor to Hive: Why not PIG?2:33 PM Jun 10th from web
- Both Hive and HBase had to struggle with Java overhead to improve performance.2:26 PM Jun 10th from web
- the columnar storage has been introduced to Hive, recently, showing better compression.2:24 PM Jun 10th from web
- Hive is also getting matured by adopting (traditional) query optimization.2:23 PM Jun 10th from web
- upcoming HBase 0.20 will show improvement on random-access latency2:05 PM Jun 10th from web
- The room for Track 1 is expanded, which is a good example of elasticity :-) #hadoopsummit091:34 PM Jun 10th from web
- ... (3) Proof-of-Concept and ad-hoc work (10%), (4) development, testing, and QA (10%).10:51 AM Jun 10th from web
- 4 tiers of Hadoop deployment in Y!: (1) production systems (20%), (2) science and research (60%), ...10:51 AM Jun 10th from web
|
- Name Jun Tatemura
- Location Cupertino, CA
|
