Featured post
mapreduce - Hadoop namenode : Single point of failure -
the namenode in hadoop architecture single point of failure.
how people have large hadoop clusters cope problem?.
is there industry-accepted solution has worked wherein secondary namenode takes on in case primary 1 fails ?
yahoo has certain recommendations configuration settings @ different cluster sizes take namenode failure account. example:
the single point of failure in hadoop cluster namenode. while loss of other machine (intermittently or permanently) not result in data loss, namenode loss results in cluster unavailability. permanent loss of namenode data render cluster's hdfs inoperable.
therefore, step should taken in configuration namenode metadata
facebook uses a tweaked version of hadoop data warehouses; has some optimizations focus on namenode reliability. additionally patches available on github, facebook appears use avatarnode switching between primary , secondary namenodes. dhruba borthakur's blog contains several other entries offering further insights namenode single point of failure.
- Get link
- X
- Other Apps
Comments
Post a Comment