When you say small files, do you mean to say those are stored within HBase columns? If so, you need not worry as HBase would eventually write bigger HFile on disk (or HDFS). If you are storing lot of small files on HDFS itself, then you will have scalability problems as single NameNode cannot handle billions of files.
2012/10/25 amit bohra <[email protected]> > Hi, > > We are working on processing of lot of small files. For processing them we > are using HBase Mapreduce as of now. Currently we are working with files in > the range for around few millions, but over the period of time it would > grow to a larger extent. > > Did anyone faced any issues while working on HBase mapreduce on small > files? > > Thanks and Regards, > Amit Bohra > -- Have a Nice Day! Lohit
