Using HDFS URLs
hdfs URL scheme when accessing files in HDFS for external tables or loading data. With this scheme, Vertica bypasses the slower, less-stable WebHDFS service when possible, and falls back to it when necessary. To use WebHDFS for all access, set the HDFSUseWebHDFS configuration parameter and continue to use
hdfs URLs; you do not need to rewrite URLs to change connection methods. (See Apache Hadoop Parameters.)
You cannot use the
hdfs scheme for communal storage in Eon Mode; you must use
webhdfs to access communal storage on HDFS. However, you can use the
hdfs scheme for reading data in Eon Mode as well as in Enterprise Mode.
Vertica requires access to certain configuration files from your HDFS cluster, whether you are using the
hdfs scheme for load access or the
webhdfs scheme for communal storage.
If a data file you want to read resides on an HDFS cluster that uses Kerberos authentication, Vertica uses the current user's principal, session doAs user, or session delegation token. See Accessing Kerberized HDFS Data for more information about these options.
For additional information, see HDFS File System (Libhdfs++).