CO: Support RLE with Zstandard

src/backend/access/common/reloptions_gp.c

@@ -33,6 +33,7 @@
 #include "utils/memutils.h"
 #include "miscadmin.h"
 #include "nodes/makefuncs.h"
+#include "storage/gp_compress.h"
 
 /*
  * Helper macro used for validation
@@ -960,12 +961,12 @@ validate_and_adjust_options(StdRdOptions *result,
 
 		if (result->compresstype[0] &&
 			(pg_strcasecmp(result->compresstype, "rle_type") == 0) &&
-			(result->compresslevel > 4))
+			(result->compresslevel > RLE_MAX_LEVEL))
 		{
 			if (validate)
 				ereport(ERROR,
 						(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
-						 errmsg("compresslevel=%d is out of range for rle_type (should be in the range 1 to 4)",
+						 errmsg("compresslevel=%d is out of range for rle_type (should be in the range 1 to 6)",
 								result->compresslevel)));
 
 			result->compresslevel = setDefaultCompressionLevel(result->compresstype);

src/backend/utils/datumstream/datumstream.c

@@ -434,7 +434,19 @@ init_datumstream_info(
 				ao_attr->compressType = "zlib";
 				ao_attr->compressLevel = 9;
 				break;
+#ifdef HAVE_LIBZSTD
+			case 5:
+				ao_attr->compress = true;
+				ao_attr->compressType = "zstd";
+				ao_attr->compressLevel = 1;
+				break;
 
+			case 6:
+				ao_attr->compress = true;
+				ao_attr->compressType = "zstd";
+				ao_attr->compressLevel = 3; /* zstd recommended default */
+				break;
+#endif
 			default:
 				ereport(ERROR,
 						(errmsg("Unexpected compresslevel %d",

src/include/storage/gp_compress.h

@@ -80,7 +80,13 @@ typedef struct
 extern void zstd_free_context(zstd_context *context);
 extern zstd_context *zstd_alloc_context(void);
 
-#endif	/* USE_ZSTD */
+#define RLE_MAX_LEVEL	(6)
+
+#else
+
+#define RLE_MAX_LEVEL	(4)
+
+#endif	/* HAVE_LIBZSTD */
 
 
 #endif

src/test/regress/expected/dsp.out

@@ -561,17 +561,16 @@ create table co6(
 	a int encoding (compresstype=rle_type),
 	b float encoding (blocksize=8192))
 	distributed by (a);
-ERROR:  compresslevel=5 is out of range for rle_type (should be in the range 1 to 4)
 create table co7(a int, b float,
 	default column encoding (compresstype=RLE_TYPE, compresslevel=7))
 	distributed by (a);
-ERROR:  compresslevel=7 is out of range for rle_type (should be in the range 1 to 4)
+ERROR:  compresslevel=7 is out of range for rle_type (should be in the range 1 to 6)
 -- negative tests - session level set
 set gp_default_storage_options = "compresstype=zlib,compresslevel=11";
 ERROR:  compresslevel=11 is out of range for zlib (should be in the range 1 to 9)
 set gp_default_storage_options = "compresslevel=5,compresstype=RLE_TYPE";
-ERROR:  compresslevel=5 is out of range for rle_type (should be in the range 1 to 4)
-set gp_default_storage_options = "compresslevel=1,compresstype=rle";
+ERROR:  rle_type cannot be used with Append Only relations row orientation
+set gp_default_storage_options="compresslevel=1,compresstype=rle";
 ERROR:  unknown compresstype "rle"
 set gp_default_storage_options = "checksum=1234";
 ERROR:  invalid bool value "1234" for storage option "checksum"

src/test/regress/expected/rle.out

@@ -11329,3 +11329,38 @@ insert into sml_rle_hdr values (-1,-1.1);
 set client_min_messages=warning;
 update sml_rle_hdr set b = b + 10 where a = -1;
 commit;
+-- Some smoke tests against rle's levels using zstd
+create table co_rle_zstd1(i int encoding(compresstype=rle_type, compresslevel=5)) with (appendonly=true, orientation=column);
+create table co_rle_zstd3(i int encoding(compresstype=rle_type, compresslevel=6)) with (appendonly=true, orientation=column);
+\d+ co_rle_zstd1
+                                    Append-Only Columnar Table "public.co_rle_zstd1"
+ Column |  Type   | Modifiers | Storage | Stats target | Compression Type | Compression Level | Block Size | Description 
+--------+---------+-----------+---------+--------------+------------------+-------------------+------------+-------------
+ i      | integer |           | plain   |              | rle_type         | 5                 | 32768      | 
+Checksum: t
+Distributed by: (i)
+Options: appendonly=true, orientation=column
+
+\d+ co_rle_zstd3
+                                    Append-Only Columnar Table "public.co_rle_zstd3"
+ Column |  Type   | Modifiers | Storage | Stats target | Compression Type | Compression Level | Block Size | Description 
+--------+---------+-----------+---------+--------------+------------------+-------------------+------------+-------------
+ i      | integer |           | plain   |              | rle_type         | 6                 | 32768      | 
+Checksum: t
+Distributed by: (i)
+Options: appendonly=true, orientation=column
+
+insert into co_rle_zstd1 select generate_series(1, 100000);
+insert into co_rle_zstd3 select generate_series(1, 100000);
+select count(distinct i) from co_rle_zstd1;
+ count  
+--------
+ 100000
+(1 row)
+
+select count(distinct i) from co_rle_zstd3;
+ count  
+--------
+ 100000
+(1 row)
+

src/test/regress/sql/rle.sql

@@ -4851,3 +4851,13 @@ insert into sml_rle_hdr values (-1,-1.1);
 set client_min_messages=warning;
 update sml_rle_hdr set b = b + 10 where a = -1;
 commit;
+
+-- Some smoke tests against rle's levels using zstd
+create table co_rle_zstd1(i int encoding(compresstype=rle_type, compresslevel=5)) with (appendonly=true, orientation=column);
+create table co_rle_zstd3(i int encoding(compresstype=rle_type, compresslevel=6)) with (appendonly=true, orientation=column);
+\d+ co_rle_zstd1
+\d+ co_rle_zstd3
+insert into co_rle_zstd1 select generate_series(1, 100000);
+insert into co_rle_zstd3 select generate_series(1, 100000);
+select count(distinct i) from co_rle_zstd1;
+select count(distinct i) from co_rle_zstd3;