Hi Ashok,
Thank you for your kind remarks. I shall endeavour to get the write-up on how to make Hive use Spark as execution engine as soon as possible. With regard to why having Hive using Spark as opposed to default MapReduce, I think we ought to look at case by case. The standard MapReduce (MR) as I know reads the data from HDFS, apply MR algorithm and writes back to HDFS. If there are many iterations of MR then, there will be many intermediate writes to HDFS. This is all serial. Spark is supposed to improve the query response by considering the whole DAG (directed acyclic graph) of map-reduce steps and optimizing it globally (e.g., pipelining consecutive map steps into one, not write intermediate data to HDFS). So in short this prevents writing data back and forth after every reduce step. However, Spark relies on the available memory. As I recall which I think it is generally applicable to RDBMS (Oracle, Sybase) and Big Data equally, a Logical IO (LIO) is around 20 times faster than a Physical IO (PIO). The first one is read from buffer cache (memory) and the latter is disk read. So hence Spark wins on that front. However, the caveat is – please correct me if I am wrong, if you do not have enough RAM to keep the intermediate data in memory, then the data has to spill to disk much like the classic Hash Joins in RDNMS. Once Spark start doing that then the advantage goes away. So in summary Spark is more performant where the underlying query results in many intermediate stages that can be fit into memory say multiple joins on many tables with smaller result set.. For such small/medium size (these being relative terms) queries, using MapReduce is like breaking a nut with sledgehammer. On the other hand for classic large batch volumes where the query response is secondary and large volumes are involved then MapReduce combines storage economy (HDFS actually) plus ab efficient computation albeit serialized. My understanding is that Tez is effectively a MapReduce engine with DAG on top. Having said that I have not used Tez personally. HTH Mich Talebzadeh Sybase ASE 15 Gold Medal Award 2008 A Winning Strategy: Running the most Critical Financial Data on ASE 15 http://login.sybase.com/files/Product_Overviews/ASE-Winning-Strategy-091908.pdf Author of the books "A Practitioner’s Guide to Upgrading to Sybase ASE 15", ISBN 978-0-9563693-0-7. co-author "Sybase Transact SQL Guidelines Best Practices", ISBN 978-0-9759693-0-4 Publications due shortly: Complex Event Processing in Heterogeneous Environments, ISBN: 978-0-9563693-3-8 Oracle and Sybase, Concepts and Contrasts, ISBN: 978-0-9563693-1-4, volume one out shortly http://talebzadehmich.wordpress.com <http://talebzadehmich.wordpress.com/> NOTE: The information in this email is proprietary and confidential. This message is for the designated recipient only, if you are not the intended recipient, you should destroy it immediately. Any information in this message shall not be understood as given or endorsed by Peridale Technology Ltd, its subsidiaries or their employees, unless expressly so stated. It is the responsibility of the recipient to ensure that this email is virus free, therefore neither Peridale Ltd, its subsidiaries nor their employees accept any responsibility. From: Ashok Kumar [mailto:ashok34...@yahoo.com] Sent: 07 December 2015 22:19 To: User <user@hive.apache.org> Subject: Fw: Managed to make Hive run on Spark engine This is great news sir. It shows perseverance pays at last. Can you inform us when the write-up is ready so I can set it up as well please. I know a bit about the advantages of having Hive using Spark engine. However, the general question I have is when one should use Hive on spark as opposed to Hive on MapReduce engine? Thanks again
