Support ZSTD with finalized format

Summary:
ZSTD 1.0.0 is coming. We can finally add a support of ZSTD without worrying about compatibility.
Still keep ZSTDNotFinal for compatibility reason.

Test Plan: Run all tests. Run db_bench with ZSTD version with RocksDB built with ZSTD 1.0 and older.

Reviewers: andrewkr, IslamAbdelRahman

Reviewed By: IslamAbdelRahman

Subscribers: cyan, igor, IslamAbdelRahman, leveldb, andrewkr, dhruba

Differential Revision: https://reviews.facebook.net/D63141

HISTORY.md

@@ -4,6 +4,7 @@
 * CancelAllBackgroundWork() flushes all memtables for databases containing writes that have bypassed the WAL (writes issued with WriteOptions::disableWAL=true) before shutting down background threads.
 * Merge options source_compaction_factor, max_grandparent_overlap_bytes and expanded_compaction_factor into max_compaction_bytes.
 * Remove ImmutableCFOptions.
+* Add a compression type ZSTD, which can work with ZSTD 0.8.0 or up. Still keep ZSTDNotFinal for compatibility reasons.
 
 ### New Features
 * Introduce NewClockCache, which is based on CLOCK algorithm with better concurrent performance in some cases. It can be used to replace the default LRU-based block cache and table cache. To use it, RocksDB need to be linked with TBB lib.

build_tools/dependencies.sh

@@ -6,7 +6,7 @@ SNAPPY_BASE=/mnt/gvfs/third-party2/snappy/8c38a4c1e52b4c2cc8a9cdc31b9c947ed7dbfc
 ZLIB_BASE=/mnt/gvfs/third-party2/zlib/0882df3713c7a84f15abe368dc004581f20b39d7/1.2.8/gcc-5-glibc-2.23/9bc6787
 BZIP2_BASE=/mnt/gvfs/third-party2/bzip2/740325875f6729f42d28deaa2147b0854f3a347e/1.0.6/gcc-5-glibc-2.23/9bc6787
 LZ4_BASE=/mnt/gvfs/third-party2/lz4/0e790b441e2d9acd68d51e1d2e028f88c6a79ddf/r131/gcc-5-glibc-2.23/9bc6787
-ZSTD_BASE=/mnt/gvfs/third-party2/zstd/bd13f49da8633897105b2b8541b106d487ce54ca/0.7.5/gcc-5-glibc-2.23/9bc6787
+ZSTD_BASE=/mnt/gvfs/third-party2/zstd/9455f75ff7f4831dc9fda02a6a0f8c68922fad8f/1.0.0/gcc-5-glibc-2.23/9bc6787
 GFLAGS_BASE=/mnt/gvfs/third-party2/gflags/f001a51b2854957676d07306ef3abf67186b5c8b/2.1.1/gcc-4.8.1-glibc-2.17/c3f970a
 JEMALLOC_BASE=/mnt/gvfs/third-party2/jemalloc/fc8a13ca1fffa4d0765c716c5a0b49f0c107518f/master/gcc-5-glibc-2.23/1c32b4b
 NUMA_BASE=/mnt/gvfs/third-party2/numa/17c514c4d102a25ca15f4558be564eeed76f4b6a/2.0.8/gcc-5-glibc-2.23/9bc6787

build_tools/dependencies_4.8.1.sh

@@ -6,7 +6,7 @@ SNAPPY_BASE=/mnt/gvfs/third-party2/snappy/8c38a4c1e52b4c2cc8a9cdc31b9c947ed7dbfc
 ZLIB_BASE=/mnt/gvfs/third-party2/zlib/0882df3713c7a84f15abe368dc004581f20b39d7/1.2.8/gcc-4.8.1-glibc-2.17/c3f970a
 BZIP2_BASE=/mnt/gvfs/third-party2/bzip2/740325875f6729f42d28deaa2147b0854f3a347e/1.0.6/gcc-4.8.1-glibc-2.17/c3f970a
 LZ4_BASE=/mnt/gvfs/third-party2/lz4/0e790b441e2d9acd68d51e1d2e028f88c6a79ddf/r131/gcc-4.8.1-glibc-2.17/c3f970a
-ZSTD_BASE=/mnt/gvfs/third-party2/zstd/bd13f49da8633897105b2b8541b106d487ce54ca/0.7.5/gcc-4.8.1-glibc-2.17/c3f970a
+ZSTD_BASE=/mnt/gvfs/third-party2/zstd/9455f75ff7f4831dc9fda02a6a0f8c68922fad8f/1.0.0/gcc-4.8.1-glibc-2.17/c3f970a
 GFLAGS_BASE=/mnt/gvfs/third-party2/gflags/f001a51b2854957676d07306ef3abf67186b5c8b/2.1.1/gcc-4.8.1-glibc-2.17/c3f970a
 JEMALLOC_BASE=/mnt/gvfs/third-party2/jemalloc/fc8a13ca1fffa4d0765c716c5a0b49f0c107518f/master/gcc-4.8.1-glibc-2.17/8d31e51
 NUMA_BASE=/mnt/gvfs/third-party2/numa/17c514c4d102a25ca15f4558be564eeed76f4b6a/2.0.8/gcc-4.8.1-glibc-2.17/c3f970a

db/db_test.cc

@@ -1193,7 +1193,7 @@ bool MinLevelToCompress(CompressionType& type, Options& options, int wbits,
     type = kXpressCompression;
     fprintf(stderr, "using xpress\n");
   } else if (ZSTD_Supported()) {
-    type = kZSTDNotFinalCompression;
+    type = kZSTD;
     fprintf(stderr, "using ZSTD\n");
   } else {
     fprintf(stderr, "skipping test, compression disabled\n");
@@ -4708,7 +4708,7 @@ TEST_F(DBTest, CompressionStatsTest) {
     type = kXpressCompression;
     fprintf(stderr, "using xpress\n");
   } else if (ZSTD_Supported()) {
-    type = kZSTDNotFinalCompression;
+    type = kZSTD;
     fprintf(stderr, "using ZSTD\n");
   } else {
     fprintf(stderr, "skipping test, compression disabled\n");

db/db_test2.cc

@@ -961,9 +961,9 @@ TEST_F(DBTest2, PresetCompressionDict) {
   compression_types.push_back(kLZ4Compression);
   compression_types.push_back(kLZ4HCCompression);
 #endif                          // LZ4_VERSION_NUMBER >= 10400
-#if ZSTD_VERSION_NUMBER >= 500  // v0.5.0+
-  compression_types.push_back(kZSTDNotFinalCompression);
-#endif  // ZSTD_VERSION_NUMBER >= 500
+  if (ZSTD_Supported()) {
+    compression_types.push_back(kZSTD);
+  }
 
   for (auto compression_type : compression_types) {
     options.compression = compression_type;

include/rocksdb/options.h

@@ -63,7 +63,13 @@ enum CompressionType : unsigned char {
   kLZ4Compression = 0x4,
   kLZ4HCCompression = 0x5,
   kXpressCompression = 0x6,
-  // zstd format is not finalized yet so it's subject to changes.
+  kZSTD = 0x7,
+
+  // Only use kZSTDNotFinalCompression if you have to use ZSTD lib older than
+  // 0.8.0 or consider a possibility of downgrading the service or copying
+  // the database files to another service running with an older version of
+  // RocksDB that doesn't have kZSTD. Otherwise, you should use kZSTD. We will
+  // eventually remove the option from the public API.
   kZSTDNotFinalCompression = 0x40,
 
   // kDisableCompressionOption is used to disable some compression options.

table/block_based_table_builder.cc

@@ -377,6 +377,7 @@ Slice CompressBlock(const Slice& raw,
         return *compressed_output;
       }
       break;
+    case kZSTD:
     case kZSTDNotFinalCompression:
       if (ZSTD_Compress(compression_options, raw.data(), raw.size(),
                         compressed_output, compression_dict) &&

table/format.cc

@@ -498,6 +498,7 @@ Status UncompressBlockContentsForCompressionType(
       *contents =
         BlockContents(std::move(ubuf), decompress_size, true, kNoCompression);
       break;
+    case kZSTD:
     case kZSTDNotFinalCompression:
       ubuf.reset(ZSTD_Uncompress(data, n, &decompress_size, compression_dict));
       if (!ubuf) {

table/table_test.cc

@@ -588,8 +588,8 @@ static std::vector<TestArgs> GenerateArgList() {
     compression_types.emplace_back(kXpressCompression, true);
   }
   if (ZSTD_Supported()) {
-    compression_types.emplace_back(kZSTDNotFinalCompression, false);
-    compression_types.emplace_back(kZSTDNotFinalCompression, true);
+    compression_types.emplace_back(kZSTD, false);
+    compression_types.emplace_back(kZSTD, true);
   }
 
   for (auto test_type : test_types) {

tools/db_bench_tool.cc

@@ -597,7 +597,7 @@ enum rocksdb::CompressionType StringToCompressionType(const char* ctype) {
   else if (!strcasecmp(ctype, "xpress"))
     return rocksdb::kXpressCompression;
   else if (!strcasecmp(ctype, "zstd"))
-    return rocksdb::kZSTDNotFinalCompression;
+    return rocksdb::kZSTD;
 
   fprintf(stdout, "Cannot parse compression type '%s'\n", ctype);
   return rocksdb::kSnappyCompression;  // default value
@@ -1771,7 +1771,7 @@ class Benchmark {
         ok = XPRESS_Compress(input.data(),
           input.size(), compressed);
         break;
-      case rocksdb::kZSTDNotFinalCompression:
+      case rocksdb::kZSTD:
         ok = ZSTD_Compress(Options().compression_opts, input.data(),
                            input.size(), compressed);
         break;
@@ -2606,7 +2606,7 @@ class Benchmark {
           &decompress_size);
         ok = uncompressed != nullptr;
         break;
-      case rocksdb::kZSTDNotFinalCompression:
+      case rocksdb::kZSTD:
         uncompressed = ZSTD_Uncompress(compressed.data(), compressed.size(),
                                        &decompress_size);
         ok = uncompressed != nullptr;

tools/db_sanity_test.cc

@@ -186,7 +186,7 @@ class SanityTestZSTDCompression : public SanityTest {
  public:
   explicit SanityTestZSTDCompression(const std::string& path)
       : SanityTest(path) {
-    options_.compression = kZSTDNotFinalCompression;
+    options_.compression = kZSTD;
   }
   virtual Options GetOptions() const override { return options_; }
   virtual std::string Name() const override { return "ZSTDCompression"; }

tools/db_stress.cc

@@ -375,7 +375,7 @@ enum rocksdb::CompressionType StringToCompressionType(const char* ctype) {
   else if (!strcasecmp(ctype, "xpress"))
     return rocksdb::kXpressCompression;
   else if (!strcasecmp(ctype, "zstd"))
-    return rocksdb::kZSTDNotFinalCompression;
+    return rocksdb::kZSTD;
 
   fprintf(stdout, "Cannot parse compression type '%s'\n", ctype);
   return rocksdb::kSnappyCompression; //default value

tools/ldb_cmd.cc

@@ -509,7 +509,7 @@ Options LDBCommand::PrepareOptionsForOpenDB() {
     } else if (comp == "xpress") {
       opt.compression = kXpressCompression;
     } else if (comp == "zstd") {
-      opt.compression = kZSTDNotFinalCompression;
+      opt.compression = kZSTD;
     } else {
       // Unknown compression.
       exec_state_ =

tools/sst_dump_tool.cc

@@ -185,16 +185,15 @@ int SstFileReader::ShowAllCompressionSizes(size_t block_size) {
 
   fprintf(stdout, "Block Size: %" ROCKSDB_PRIszt "\n", block_size);
 
-  std::pair<CompressionType,const char*> compressions[] = {
-    { CompressionType::kNoCompression, "kNoCompression" },
-    { CompressionType::kSnappyCompression, "kSnappyCompression" },
-    { CompressionType::kZlibCompression, "kZlibCompression" },
-    { CompressionType::kBZip2Compression, "kBZip2Compression" },
-    { CompressionType::kLZ4Compression, "kLZ4Compression" },
-    { CompressionType::kLZ4HCCompression, "kLZ4HCCompression" },
-    { CompressionType::kXpressCompression, "kXpressCompression" },
-    { CompressionType::kZSTDNotFinalCompression, "kZSTDNotFinalCompression" }
-  };
+  std::pair<CompressionType, const char*> compressions[] = {
+      {CompressionType::kNoCompression, "kNoCompression"},
+      {CompressionType::kSnappyCompression, "kSnappyCompression"},
+      {CompressionType::kZlibCompression, "kZlibCompression"},
+      {CompressionType::kBZip2Compression, "kBZip2Compression"},
+      {CompressionType::kLZ4Compression, "kLZ4Compression"},
+      {CompressionType::kLZ4HCCompression, "kLZ4HCCompression"},
+      {CompressionType::kXpressCompression, "kXpressCompression"},
+      {CompressionType::kZSTD, "kZSTD"}};
 
   for (auto& i : compressions) {
     if (CompressionTypeSupported(i.first)) {

util/compression.h

@@ -79,6 +79,14 @@ inline bool XPRESS_Supported() {
 }
 
 inline bool ZSTD_Supported() {
+#ifdef ZSTD
+  // ZSTD format is finalized since version 0.8.0.
+  return (ZSTD_versionNumber() >= 800);
+#endif
+  return false;
+}
+
+inline bool ZSTDNotFinal_Supported() {
 #ifdef ZSTD
   return true;
 #endif
@@ -102,6 +110,8 @@ inline bool CompressionTypeSupported(CompressionType compression_type) {
     case kXpressCompression:
       return XPRESS_Supported();
     case kZSTDNotFinalCompression:
+      return ZSTDNotFinal_Supported();
+    case kZSTD:
       return ZSTD_Supported();
     default:
       assert(false);
@@ -125,6 +135,7 @@ inline std::string CompressionTypeToString(CompressionType compression_type) {
       return "LZ4HC";
     case kXpressCompression:
       return "Xpress";
+    case kZSTD:
     case kZSTDNotFinalCompression:
       return "ZSTD";
     default:

util/options_helper.h

@@ -598,6 +598,7 @@ static std::unordered_map<std::string, CompressionType>
         {"kLZ4Compression", kLZ4Compression},
         {"kLZ4HCCompression", kLZ4HCCompression},
         {"kXpressCompression", kXpressCompression},
+        {"kZSTD", kZSTD},
         {"kZSTDNotFinalCompression", kZSTDNotFinalCompression},
         {"kDisableCompressionOption", kDisableCompressionOption}};
 

util/options_test.cc

@@ -56,6 +56,7 @@ TEST_F(OptionsTest, GetOptionsFromMapTest) {
        "kLZ4Compression:"
        "kLZ4HCCompression:"
        "kXpressCompression:"
+       "kZSTD:"
        "kZSTDNotFinalCompression"},
       {"bottommost_compression", "kLZ4Compression"},
       {"compression_opts", "4:5:6:7"},
@@ -140,7 +141,7 @@ TEST_F(OptionsTest, GetOptionsFromMapTest) {
   ASSERT_EQ(new_cf_opt.min_write_buffer_number_to_merge, 3);
   ASSERT_EQ(new_cf_opt.max_write_buffer_number_to_maintain, 99);
   ASSERT_EQ(new_cf_opt.compression, kSnappyCompression);
-  ASSERT_EQ(new_cf_opt.compression_per_level.size(), 8U);
+  ASSERT_EQ(new_cf_opt.compression_per_level.size(), 9U);
   ASSERT_EQ(new_cf_opt.compression_per_level[0], kNoCompression);
   ASSERT_EQ(new_cf_opt.compression_per_level[1], kSnappyCompression);
   ASSERT_EQ(new_cf_opt.compression_per_level[2], kZlibCompression);
@@ -148,7 +149,8 @@ TEST_F(OptionsTest, GetOptionsFromMapTest) {
   ASSERT_EQ(new_cf_opt.compression_per_level[4], kLZ4Compression);
   ASSERT_EQ(new_cf_opt.compression_per_level[5], kLZ4HCCompression);
   ASSERT_EQ(new_cf_opt.compression_per_level[6], kXpressCompression);
-  ASSERT_EQ(new_cf_opt.compression_per_level[7], kZSTDNotFinalCompression);
+  ASSERT_EQ(new_cf_opt.compression_per_level[7], kZSTD);
+  ASSERT_EQ(new_cf_opt.compression_per_level[8], kZSTDNotFinalCompression);
   ASSERT_EQ(new_cf_opt.compression_opts.window_bits, 4);
   ASSERT_EQ(new_cf_opt.compression_opts.level, 5);
   ASSERT_EQ(new_cf_opt.compression_opts.strategy, 6);

utilities/column_aware_encoding_exp.cc

@@ -80,8 +80,7 @@ class ColumnAwareEncodingExp {
     std::unordered_map<std::string, CompressionType> compressions = {
         {"kNoCompression", CompressionType::kNoCompression},
         {"kZlibCompression", CompressionType::kZlibCompression},
-        {"kZSTDNotFinalCompression",
-         CompressionType::kZSTDNotFinalCompression}};
+        {"kZSTD", CompressionType::kZSTD}};
 
     // Find Compression
     CompressionType compression_type = compressions[FLAGS_compression_type];