enabled encoding and dictionary

Author: nikhilsinhaparseable

Diff for: src/storage/object_storage.rs

@@ -21,7 +21,9 @@ use super::{
     ObjectStoreFormat, Permisssion, StorageDir, StorageMetadata,
 };
 use super::{
-    LogStream, Owner, StreamType, ALERTS_ROOT_DIRECTORY, MANIFEST_FILE, PARSEABLE_METADATA_FILE_NAME, PARSEABLE_ROOT_DIRECTORY, SCHEMA_FILE_NAME, STREAM_METADATA_FILE_NAME, STREAM_ROOT_DIRECTORY
+    LogStream, Owner, StreamType, ALERTS_ROOT_DIRECTORY, MANIFEST_FILE,
+    PARSEABLE_METADATA_FILE_NAME, PARSEABLE_ROOT_DIRECTORY, SCHEMA_FILE_NAME,
+    STREAM_METADATA_FILE_NAME, STREAM_ROOT_DIRECTORY,
 };
 
 use crate::alerts::AlertConfig;

Diff for: src/storage/staging.rs

@@ -339,12 +339,31 @@ pub fn parquet_writer_props(
         sorting_column_vec.push(sorting_column);
     }
 
+    props = props
+        .set_dictionary_enabled(true)
+        .set_encoding(Encoding::PLAIN)
+        .set_statistics_enabled(EnabledStatistics::Chunk);
+
     let url_column = ColumnPath::new(vec!["URL".to_string()]);
     props = props
         .set_column_dictionary_enabled(url_column.clone(), true)
         .set_column_encoding(url_column.clone(), Encoding::DELTA_BYTE_ARRAY)
         .set_column_statistics_enabled(url_column.clone(), EnabledStatistics::Chunk);
 
+    let event_time_column = ColumnPath::new(vec!["EventTime".to_string()]);
+    props = props
+        .set_column_encoding(event_time_column.clone(), Encoding::DELTA_BINARY_PACKED)
+        .set_column_statistics_enabled(event_time_column.clone(), EnabledStatistics::Chunk);
+
+    let user_id_column = ColumnPath::new(vec!["UserID".to_string()]);
+    props = props
+        .set_column_encoding(user_id_column.clone(), Encoding::RLE)
+        .set_column_statistics_enabled(user_id_column.clone(), EnabledStatistics::Chunk);
+
+    let search_phrase_column = ColumnPath::new(vec!["SearchPhrase".to_string()]);
+    props = props
+        .set_column_encoding(search_phrase_column.clone(), Encoding::DELTA_BYTE_ARRAY)
+        .set_column_statistics_enabled(search_phrase_column.clone(), EnabledStatistics::Chunk);
     props = props.set_sorting_columns(Some(sorting_column_vec));
     props
 }