hadoop-common-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Bennett Andrews <bennett.j.andr...@gmail.com>
Subject Re: HBASE on Hadoop
Date Thu, 28 Apr 2011 19:25:37 GMT
Setup a "distributed" HBase cluster.

On each node that runs TaskTracker and Datanode, also run a HBase
RegionServer.

On the node that runs JobTracker and Namenode, run the HBase Master.

There is a getting started guide here.
http://hbase.apache.org/book/notsoquick.html

Check out the HBase user list for more.
http://hbase.apache.org/mail-lists.html



On Wed, Apr 27, 2011 at 4:58 PM, gaurav garg <gaurav.garg28@gmail.com>wrote:

> Praveenesh,
>
> I will recommend you to read the google Big Table paper(
> http://labs.google.com/papers/bigtable.html) which is a foundation for the
> hbase.
> Terminology is little different though:
>
> Mapping of terms(not exhaustive):
> ***********************************
> Big Table             Hbase
> ***********************************
> Master Server      HMaster
> Tablet                  region
> Tablet server        regionserver
>  chubby                zookeeper (It is apache implementation of
> distributed
> synchronization server)
>
> Hbase stores data on hadoop dfs. hbase is a client of hdfs. Hence hadoop
> will automatically distribute and replicate the data across your hadoop
> cluster.
> Hbase master and regionservers formats/transforms the data and relies on
> hadoop for storage and retrieval.
>
>
> Hbase cluster can run on a separate set of nodes or it can even share
> hadoop
> nodes.
>
> Once you have setup hdfs cluster, hbase cluster can be easily setup.
>
> Thanks
> Gaurav
>
>
> On Tue, Apr 26, 2011 at 10:55 AM, praveenesh kumar <praveenesh@gmail.com
> >wrote:
>
> > Hello everyone,
> >
> > Thanks everyone for guiding me everytime. I am able to setup hadoop
> cluster
> > of 10 nodes.
> > Now comes HBASE..!!!
> >
> > I am new to all this...
> > My problem is I have huge data to analyze.
> > so shall I go for single node Hbase installation on all nodes or go for
> > distributed Hbase installation.??
> >
> > How distributed installation is different from single node installaion ??
> > Now suppose if I have distributed Hbase...
> > and If I design some table on my master node.. and then store data on
> it..
> > say around 100M. How the data is going to be distributed.. Will HBASE do
> it
> > automatically or we have to write codes for getting it distributed ??
> > Is there any good tutorial that tells us more about HBase and how to work
> > on
> > it ???
> >
> > Thanks,
> > Praveenesh
> >
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message