0

I am successfully saving a complex object (scala.Vector[(Int, Double)]) in an HBase table column by means of serializing it to a byte array (scala.Array[Byte]) with Kryo (com.twitter.chill.KryoInjection).

Now I would like to query this data with Apache Drill and convert it from the serialized byte array to something human readable.

I imagined I would somehow "register" my own converter with Apache Drill and provide it to drill query.

Unfortunately I was unable to find any information on if and how it is possible to accomplish.

Reinis
  • 65
  • 9

1 Answers1

0

If you encode your data in big endian in hbase you will get meaningful results. You will also get better performance on conditional queries. I think this doc helps describe the problem/solution.

catpaws
  • 2,263
  • 16
  • 18
  • I am not sure if my explanation is not clear enough but I don't see what encoding has to do with the conversion by means of `serialization/de-serialization`. Additionally, I am pretty sure that written `byte array` is `big endian` encoded. – Reinis Sep 22 '15 at 16:12
  • I misunderstood the question, sorry. I thought you had the data in hbase and were using the hbase storage plugin to connect Drill, but getting unreadable results from queries. BTW, the new Drill hbase query capability described in the referenced doc is for Drill 1.2, which will be released soon. I think these features are merged in the source code already. – catpaws Sep 23 '15 at 12:10