I listened to this video link posted in Cloudera - Big Data distribution company. It gives fairly vague idea about Hbase.
Since HDFS has the drawback of sequential access, Hbase is built to overcome this drawback and to make use of random access along with HDFS. Hbase is a combination of HDFS and Random access technology together.
Please find the link for the video here:
Introduction to HBase - Todd Lipcon (Cloudera)