[WIP] Delete duplicated key/value pairs recursively

This is to implement the idea: http://pad.ceph.com/p/rocksdb-wal-improvement Add a new flush style called kFlushStyleDedup which users can config by setting flush_style=kFlushStyleDedup and write_buffer_number_to_flush. Flush is triggered while number of non-flushed memtables is >= min_write_buffer_number_to_merge. When flushing memtables to L0, it merges write_buffer_number_to_flush memtables and deduping against other non-flushed memtables. With bluestore fio plugin to do 4k random write test, I compared the two stats db.get.micros and db.write.micros from rocksdb every ten minutes. db.get.micros is the average of get operations, and db.write.micros is the average time of put/delete functions. From the result, rocksdb read performance can be improved up to 38%, and write performance can be improved to 15%. More detailed info please see the excel: https://drive.google.com/drive/folders/0B6jqFc7e2yxVdUQ2aEpCR3ItbG8 And to complement, the io perfermance of bluestore fio doesn't improve much. From time of txc states, the time spends a lot in kv_state_commiting_lat (up to 80%), but submit_transaction and submit_transaction_sync takes little time compared. Signed-off-by: Xiaoyan Li [email protected]
ceph · Aug 22, 2017 · bf29157 · bf29157
1 parent 2e64f45
commit bf29157
Show file tree

Hide file tree

Showing 27 changed files with 249 additions and 31 deletions.
diff --git a/db/builder.cc b/db/builder.cc
@@ -76,13 +76,15 @@ Status BuildTable(
     const CompressionOptions& compression_opts, bool paranoid_file_checks,
     InternalStats* internal_stats, TableFileCreationReason reason,
     EventLogger* event_logger, int job_id, const Env::IOPriority io_priority,
-    TableProperties* table_properties, int level) {
+    TableProperties* table_properties, int level,
+    InternalIterator* compare_iter) {
   assert((column_family_id ==
           TablePropertiesCollectorFactory::Context::kUnknownColumnFamily) ==
          column_family_name.empty());
   // Reports the IOStats for flush for every following bytes.
   const size_t kReportFlushIOStatsEvery = 1048576;
   Status s;
+  uint64_t num_duplicated = 0, num_total = 0;
   meta->fd.file_size = 0;
   iter->SeekToFirst();
   std::unique_ptr<RangeDelAggregator> range_del_agg(
@@ -138,17 +140,56 @@ Status BuildTable(
         &snapshots, earliest_write_conflict_snapshot, env,
         true /* internal key corruption is not ok */, range_del_agg.get());
     c_iter.SeekToFirst();
+
+    ParsedInternalKey c_ikey, comp_ikey;
+    if (compare_iter != nullptr) {
+      compare_iter->SeekToFirst();  //find the first one
+    }
+    const rocksdb::Comparator *comp = internal_comparator.user_comparator();
     for (; c_iter.Valid(); c_iter.Next()) {
       const Slice& key = c_iter.key();
       const Slice& value = c_iter.value();
-      builder->Add(key, value);
-      meta->UpdateBoundaries(key, c_iter.ikey().sequence);
-
-      // TODO(noetzli): Update stats after flush, too.
-      if (io_priority == Env::IO_HIGH &&
-          IOSTATS(bytes_written) >= kReportFlushIOStatsEvery) {
-        ThreadStatusUtil::SetThreadOperationProperty(
-            ThreadStatus::FLUSH_BYTES_WRITTEN, IOSTATS(bytes_written));
+      bool skip = false;
+      num_total++;
+
+      // Check whether the key is duplicated in later mems
+      c_ikey = c_iter.ikey();
+      if (compare_iter != nullptr ) {
+        while (compare_iter->Valid()) {
+          if (!ParseInternalKey(compare_iter->key(), &comp_ikey)) {
+            compare_iter->Next();
+            continue;
+          }
+          int result = comp->Compare(c_ikey.user_key, comp_ikey.user_key);
+          if (result == 0) {
+            // No delete for merge options.
+            if (c_ikey.type != kTypeMerge && comp_ikey.type != kTypeMerge) {
+              skip = true;
+            }
+            break;
+          }
+          else if (result > 0) {
+            compare_iter->Next();
+          }
+          else {
+            break;
+          }
+        }
+      }
+      if (skip) {
+          num_duplicated++;
+          continue;
+      }
+      else {
+          builder->Add(key, value);
+          meta->UpdateBoundaries(key, c_iter.ikey().sequence);
+
+          // TODO(noetzli): Update stats after flush, too.
+          if (io_priority == Env::IO_HIGH &&
+              IOSTATS(bytes_written) >= kReportFlushIOStatsEvery) {
+            ThreadStatusUtil::SetThreadOperationProperty(
+                ThreadStatus::FLUSH_BYTES_WRITTEN, IOSTATS(bytes_written));
+          }
       }
     }
     // nullptr for table_{min,max} so all range tombstones will be flushed
@@ -218,9 +259,10 @@ Status BuildTable(
   }
 
   // Output to event logger and fire events.
+  bool show_num = (compare_iter != nullptr)?true:false;
   EventHelpers::LogAndNotifyTableFileCreationFinished(
       event_logger, ioptions.listeners, dbname, column_family_name, fname,
-      job_id, meta->fd, tp, reason, s);
+      job_id, meta->fd, tp, reason, s, show_num, num_total, num_duplicated);
 
   return s;
 }

diff --git a/db/builder.h b/db/builder.h
@@ -79,6 +79,7 @@ extern Status BuildTable(
     InternalStats* internal_stats, TableFileCreationReason reason,
     EventLogger* event_logger = nullptr, int job_id = 0,
     const Env::IOPriority io_priority = Env::IO_HIGH,
-    TableProperties* table_properties = nullptr, int level = -1);
+    TableProperties* table_properties = nullptr, int level = -1,
+    InternalIterator* compare_iter = nullptr);
 
 }  // namespace rocksdb
diff --git a/db/c.cc b/db/c.cc
@@ -2263,6 +2263,14 @@ void rocksdb_options_set_max_write_buffer_number(rocksdb_options_t* opt, int n)
   opt->rep.max_write_buffer_number = n;
 }
 
+void rocksdb_options_set_flush_style(rocksdb_options_t* opt, int style) {
+  opt->rep.flush_style = static_cast<rocksdb::FlushStyle>(style);
+}
+
+void rocksdb_options_set_write_buffer_number_to_flush(rocksdb_options_t* opt, int n) {
+  opt->rep.write_buffer_number_to_flush = n;
+}
+
 void rocksdb_options_set_min_write_buffer_number_to_merge(rocksdb_options_t* opt, int n) {
   opt->rep.min_write_buffer_number_to_merge = n;
 }

diff --git a/db/column_family.cc b/db/column_family.cc
@@ -165,6 +165,14 @@ ColumnFamilyOptions SanitizeOptions(const ImmutableDBOptions& db_options,
     result.min_write_buffer_number_to_merge = 1;
   }
 
+  if (result.flush_style == kFlushStyleDedup) {
+    if (result.write_buffer_number_to_flush < 1 ||
+        (result.write_buffer_number_to_flush >
+         result.min_write_buffer_number_to_merge)) {
+      result.write_buffer_number_to_flush = 1;
+    }
+  }
+
   if (result.num_levels < 1) {
     result.num_levels = 1;
   }
@@ -181,6 +189,7 @@ ColumnFamilyOptions SanitizeOptions(const ImmutableDBOptions& db_options,
   if (result.max_write_buffer_number < 2) {
     result.max_write_buffer_number = 2;
   }
+
   if (result.max_write_buffer_number_to_maintain < 0) {
     result.max_write_buffer_number_to_maintain = result.max_write_buffer_number;
   }
@@ -359,7 +368,9 @@ ColumnFamilyData::ColumnFamilyData(
       write_buffer_manager_(write_buffer_manager),
       mem_(nullptr),
       imm_(ioptions_.min_write_buffer_number_to_merge,
-           ioptions_.max_write_buffer_number_to_maintain),
+           ioptions_.max_write_buffer_number_to_maintain,
+           ioptions_.write_buffer_number_to_flush),
+      stop(false),
       super_version_(nullptr),
       super_version_number_(0),
       local_sv_(new ThreadLocalPtr(&SuperVersionUnrefHandle)),

diff --git a/db/column_family.h b/db/column_family.h
@@ -236,6 +236,9 @@ class ColumnFamilyData {
   MemTable* mem() { return mem_; }
   Version* current() { return current_; }
   Version* dummy_versions() { return dummy_versions_; }
+  void set_stop() { stop = true; }
+  bool is_stop() { return stop; }
+  bool is_flush_recursive_dedup() { return (ioptions_.flush_style == kFlushStyleDedup); }
   void SetCurrent(Version* _current);
   uint64_t GetNumLiveVersions() const;  // REQUIRE: DB mutex held
   uint64_t GetTotalSstFilesSize() const;  // REQUIRE: DB mutex held
@@ -371,6 +374,7 @@ class ColumnFamilyData {
 
   MemTable* mem_;
   MemTableList imm_;
+  bool stop;
   SuperVersion* super_version_;
 
   // An ordinal representing the current SuperVersion. Updated by

diff --git a/db/column_family_test.cc b/db/column_family_test.cc
@@ -1027,6 +1027,7 @@ TEST_F(ColumnFamilyTest, DifferentWriteBufferSizes) {
   default_cf.arena_block_size = 4 * 4096;
   default_cf.max_write_buffer_number = 10;
   default_cf.min_write_buffer_number_to_merge = 1;
+  default_cf.write_buffer_number_to_flush = 0;
   default_cf.max_write_buffer_number_to_maintain = 0;
   one.write_buffer_size = 200000;
   one.arena_block_size = 4 * 4096;

diff --git a/db/compaction_picker_test.cc b/db/compaction_picker_test.cc
@@ -70,7 +70,7 @@ class CompactionPickerTest : public testing::Test {
     DeleteVersionStorage();
     options_.num_levels = num_levels;
     vstorage_.reset(new VersionStorageInfo(&icmp_, ucmp_, options_.num_levels,
-                                           style, nullptr, false));
+                                           kFlushStyleMerge, style, nullptr, false));
     vstorage_->CalculateBaseBytes(ioptions_, mutable_cf_options_);
   }
 

diff --git a/db/db_impl.cc b/db/db_impl.cc
@@ -226,6 +226,7 @@ void DBImpl::CancelAllBackgroundWork(bool wait) {
     for (auto cfd : *versions_->GetColumnFamilySet()) {
       if (!cfd->IsDropped() && !cfd->mem()->IsEmpty()) {
         cfd->Ref();
+        cfd->set_stop();
         mutex_.Unlock();
         FlushMemTable(cfd, FlushOptions());
         mutex_.Lock();

diff --git a/db/event_helpers.cc b/db/event_helpers.cc
@@ -45,7 +45,7 @@ void EventHelpers::LogAndNotifyTableFileCreationFinished(
     const std::string& db_name, const std::string& cf_name,
     const std::string& file_path, int job_id, const FileDescriptor& fd,
     const TableProperties& table_properties, TableFileCreationReason reason,
-    const Status& s) {
+    const Status& s, bool show_num, int total_num, int dup_num) {
   if (s.ok() && event_logger) {
     JSONWriter jwriter;
     AppendCurrentTime(&jwriter);
@@ -73,6 +73,9 @@ void EventHelpers::LogAndNotifyTableFileCreationFinished(
               << "num_data_blocks" << table_properties.num_data_blocks
               << "num_entries" << table_properties.num_entries
               << "filter_policy_name" << table_properties.filter_policy_name;
+      if (show_num) {
+        jwriter << "total_paris" << total_num << "duplicated_pairs" << dup_num;
+      }
 
       // user collected properties
       for (const auto& prop : table_properties.readable_properties) {

diff --git a/db/event_helpers.h b/db/event_helpers.h
@@ -33,7 +33,7 @@ class EventHelpers {
       const std::string& db_name, const std::string& cf_name,
       const std::string& file_path, int job_id, const FileDescriptor& fd,
       const TableProperties& table_properties, TableFileCreationReason reason,
-      const Status& s);
+      const Status& s, bool show_num = false, int total_num = 0, int duplicated_num = 0);
   static void LogAndNotifyTableFileDeletion(
       EventLogger* event_logger, int job_id,
       uint64_t file_number, const std::string& file_path,

diff --git a/db/flush_job.cc b/db/flush_job.cc
@@ -129,11 +129,16 @@ void FlushJob::PickMemTable() {
   assert(!pick_memtable_called);
   pick_memtable_called = true;
   // Save the contents of the earliest memtable as a new Table
-  cfd_->imm()->PickMemtablesToFlush(&mems_);
+
+  if (cfd_->is_flush_recursive_dedup() && !cfd_->is_stop()) {
+    cfd_->imm()->PickMemtablesToFlush(&mems_, &compare_mems_);
+  }
+  else {
+    cfd_->imm()->PickMemtablesToFlush(&mems_);
+  }
   if (mems_.empty()) {
     return;
   }
-
   ReportFlushInputSize(mems_);
 
   // entries mems are (implicitly) sorted in ascending order by their created
@@ -257,6 +262,7 @@ Status FlushJob::WriteLevel0Table() {
     ReadOptions ro;
     ro.total_order_seek = true;
     Arena arena;
+    Arena arena1;
     uint64_t total_num_entries = 0, total_num_deletes = 0;
     size_t total_memory_usage = 0;
     for (MemTable* m : mems_) {
@@ -274,17 +280,36 @@ Status FlushJob::WriteLevel0Table() {
       total_memory_usage += m->ApproximateMemoryUsage();
     }
 
+    std::vector<InternalIterator*> compare_memtables;
+    int comp_entries = 0;
+    for (MemTable *m: compare_mems_) {
+      compare_memtables.push_back(m->NewIterator(ro, &arena1));
+      comp_entries += m->num_entries();
+    }
+
     event_logger_->Log() << "job" << job_context_->job_id << "event"
                          << "flush_started"
                          << "num_memtables" << mems_.size() << "num_entries"
                          << total_num_entries << "num_deletes"
                          << total_num_deletes << "memory_usage"
-                         << total_memory_usage;
+                         << total_memory_usage << "mems table count "
+                         << mems_.size() << " compare mems count "
+                         << compare_mems_.size() << " compare mems total "
+                         << comp_entries;
 
     {
       ScopedArenaIterator iter(
           NewMergingIterator(&cfd_->internal_comparator(), &memtables[0],
                              static_cast<int>(memtables.size()), &arena));
+      InternalIterator *compare_iter = nullptr;
+      ScopedArenaIterator compare_scope_iter;
+      if (compare_mems_.size() > 0) {
+          compare_scope_iter.set(
+             NewMergingIterator(&cfd_->internal_comparator(), &compare_memtables[0],
+                                static_cast<int>(compare_memtables.size()), &arena1));
+          compare_iter = compare_scope_iter.get();
+      }
+
       std::unique_ptr<InternalIterator> range_del_iter(NewMergingIterator(
           &cfd_->internal_comparator(),
           range_del_iters.empty() ? nullptr : &range_del_iters[0],
@@ -308,7 +333,7 @@ Status FlushJob::WriteLevel0Table() {
           cfd_->ioptions()->compression_opts,
           mutable_cf_options_.paranoid_file_checks, cfd_->internal_stats(),
           TableFileCreationReason::kFlush, event_logger_, job_context_->job_id,
-          Env::IO_HIGH, &table_properties_, 0 /* level */);
+          Env::IO_HIGH, &table_properties_, 0, compare_iter);
       LogFlush(db_options_.info_log);
     }
     ROCKS_LOG_INFO(db_options_.info_log,

diff --git a/db/flush_job.h b/db/flush_job.h
@@ -104,6 +104,7 @@ class FlushJob {
   // Variables below are set by PickMemTable():
   FileMetaData meta_;
   autovector<MemTable*> mems_;
+  autovector<MemTable*> compare_mems_;
   VersionEdit* edit_;
   Version* base_;
   bool pick_memtable_called;

diff --git a/db/memtable_list.cc b/db/memtable_list.cc
@@ -279,6 +279,33 @@ bool MemTableList::IsFlushPending() const {
   return false;
 }
 
+// Returns the memtables that need to be flushed.
+void MemTableList::PickMemtablesToFlush(autovector<MemTable*>* ret,
+                                        autovector<MemTable*>* compare_ret) {
+  AutoThreadOperationStageUpdater stage_updater(
+      ThreadStatus::STAGE_PICK_MEMTABLES_TO_FLUSH);
+  assert(write_buffer_number_to_flush_ > 0);
+  const auto& memlist = current_->memlist_;
+  for (auto it = memlist.rbegin(); it != memlist.rend(); ++it) {
+    MemTable* m = *it;
+    if (!m->flush_in_progress_) {
+      assert(!m->flush_completed_);
+      if (ret->size() < write_buffer_number_to_flush_) {
+        num_flush_not_started_--;
+        if (num_flush_not_started_ == 0) {
+          imm_flush_needed.store(false, std::memory_order_release);
+        }
+        m->flush_in_progress_ = true;  // flushing will start very soon
+        ret->push_back(m);
+      }
+      else {
+        compare_ret->push_back(m);
+      }
+    }
+  }
+  flush_requested_ = false;  // start-flush request is complete
+}
+
 // Returns the memtables that need to be flushed.
 void MemTableList::PickMemtablesToFlush(autovector<MemTable*>* ret) {
   AutoThreadOperationStageUpdater stage_updater(

diff --git a/db/memtable_list.h b/db/memtable_list.h
@@ -158,9 +158,11 @@ class MemTableList {
  public:
   // A list of memtables.
   explicit MemTableList(int min_write_buffer_number_to_merge,
-                        int max_write_buffer_number_to_maintain)
+                        int max_write_buffer_number_to_maintain,
+                        int write_buffer_number_to_flush = 1)
       : imm_flush_needed(false),
         min_write_buffer_number_to_merge_(min_write_buffer_number_to_merge),
+        write_buffer_number_to_flush_(write_buffer_number_to_flush),
         current_(new MemTableListVersion(&current_memory_usage_,
                                          max_write_buffer_number_to_maintain)),
         num_flush_not_started_(0),
@@ -194,6 +196,7 @@ class MemTableList {
 
   // Returns the earliest memtables that needs to be flushed. The returned
   // memtables are guaranteed to be in the ascending order of created time.
+  void PickMemtablesToFlush(autovector<MemTable*>* mems, autovector<MemTable*>* compare_mems);
   void PickMemtablesToFlush(autovector<MemTable*>* mems);
 
   // Reset status of the given memtable list back to pending state so that
@@ -242,6 +245,8 @@ class MemTableList {
 
   const int min_write_buffer_number_to_merge_;
 
+  const unsigned int write_buffer_number_to_flush_;
+
   MemTableListVersion* current_;
 
   // the number of elements that still need flushing