|
Hi, We are evaluation MapR, and we are interested in knowing the improvements done for Hive/HBase by Mapr. |
|
MapR is a complete distribution for Apache Hadoop including HBase. The improvements to using HBase with MapR come from the MapR filesystem and MapR control system. For e.g. you will get: 1. Better performance 2. Mirroring and point-in-time snapshots to protect your data 3. High availability of services since there are no single-point-of-failures 4. NFS access to HBase log files 5. Alarms, space-usage tracking, and heatmap for monitoring of the cluster See http://www.mapr.com/products/only-with-mapr.html for more examples. 1
Same answer goes for Hive. Both Hive and Hbase run with no code changes on MapR. Even without any special code, they run considerably faster. Realistic HBase work-loads have demonstrated a speedup of MapR relative to stock Hadoop of about 4-6x.
(05 Jul '11, 01:16)
TedDunning ♦♦
Thanks Arvind and Ted for your replies. Appreciate your help!
(06 Jul '11, 02:26)
ghousia
|