

Aster Unleashed - Installing the Analytic Libraries
With your own Aster cluster installed and running (Getting Started with Aster Express), and a few Aster nPath examples under your belt (On the Road to nPath), it's time to unleash the full analytic power of Aster and install the remaining SQL-MR libraries. These Aster Analytic modules are a powerful suite of reusable SQL-MapReduce® functions that deliver advanced analytics on Big Data with immediate business impact by leveraging the power of MapReduce programming through standard SQL.
This will be a quick and straightforward administrative process. These libraries can only be installed on a running Aster cluster, which is why we've saved this step for now. First logon to the Queen node, either with a remote ssh session, or directly on to the Queen node and open a terminal window.
At the terminal window command prompt, change to the 'demo' directory and unzip the Analytics_Foundation.zip file. In this example, I'm creating a new target directory named 'sqlmr'. This is only needed temporarily and can be deleted once installed to our Aster cluster.
cd demo unzip Analytics_Foundation.zip -d sqlmr
A quick look at the files in this 'sqlmr' directory will give you a preview of the SQL-MR modules that we'll be loading.
Now we're ready to install these modules using the Aster ACT tool. First change directory to where we unzipped the files, then run 'act':
cd sqlmr act
ACT also has a 'cd' function, but I find it easier to do this prior to launching ACT. Now start ACT, remember the default user/password is beehive/beehive, and then issue the install command, \i, and use the install.sql script in our library directory:
\i install.sql
This may take a few minutes, depending on your PC resources. Once is completes, you can now check on the install modules using the ACT \dF command:
\dF
That's it. You now have a loaded Aster Analytics tool ready to start crunching your Big Data! Stay tuned for more articles that will demonstrate how to use these SQL-MR libraries.
Have fun!
Hi
I have written a huge SQL - MR functionality that is causing JAVA heap space issue.
com.asterdata.ncluster.util.NClusterException: Ran out of memory retrieving query results.
Now I wanted to ask can we change the heap size of each worker to handle large amount of data
since my MR is reducing Query execution time of about 3 hrs
Any suggestion is very help full at this stage
Thanks in Advance
Saumik Chuarasia