Read from blob cache first when MultiGetBlob() (#10225)

Summary:
There is currently no caching mechanism for blobs, which is not ideal especially when the database resides on remote storage (where we cannot rely on the OS page cache). As part of this task, we would like to make it possible for the application to configure a blob cache.

Pull Request resolved: https://github.com/facebook/rocksdb/pull/10225

Test Plan:
Add test cases for MultiGetBlob
In this task, we added the new API MultiGetBlob() for BlobSource.

This PR is a part of https://github.com/facebook/rocksdb/issues/10156

Reviewed By: ltamasi

Differential Revision: D37358364

Pulled By: gangliao

fbshipit-source-id: aff053a37615d96d768fb9aedde17da5618c7ae6

db/blob/blob_source.cc

@@ -10,6 +10,7 @@
 
 #include "db/blob/blob_file_reader.h"
 #include "options/cf_options.h"
+#include "table/multiget_context.h"
 
 namespace ROCKSDB_NAMESPACE {
 
@@ -152,6 +153,139 @@ Status BlobSource::GetBlob(const ReadOptions& read_options,
   return s;
 }
 
+void BlobSource::MultiGetBlob(
+    const ReadOptions& read_options,
+    const autovector<std::reference_wrapper<const Slice>>& user_keys,
+    uint64_t file_number, uint64_t file_size,
+    const autovector<uint64_t>& offsets,
+    const autovector<uint64_t>& value_sizes, autovector<Status*>& statuses,
+    autovector<PinnableSlice*>& blobs, uint64_t* bytes_read) {
+  size_t num_blobs = user_keys.size();
+  assert(num_blobs > 0);
+  assert(num_blobs <= MultiGetContext::MAX_BATCH_SIZE);
+  assert(num_blobs == offsets.size());
+  assert(num_blobs == value_sizes.size());
+  assert(num_blobs == statuses.size());
+  assert(num_blobs == blobs.size());
+
+#ifndef NDEBUG
+  for (size_t i = 0; i < offsets.size() - 1; ++i) {
+    assert(offsets[i] <= offsets[i + 1]);
+  }
+#endif  // !NDEBUG
+
+  using Mask = uint64_t;
+  Mask cache_hit_mask = 0;
+
+  Status s;
+  uint64_t total_bytes = 0;
+  const OffsetableCacheKey base_cache_key(db_id_, db_session_id_, file_number,
+                                          file_size);
+
+  if (blob_cache_) {
+    size_t cached_blob_count = 0;
+    for (size_t i = 0; i < num_blobs; ++i) {
+      CachableEntry<std::string> blob_entry;
+      const CacheKey cache_key = base_cache_key.WithOffset(offsets[i]);
+      const Slice key = cache_key.AsSlice();
+
+      s = GetBlobFromCache(key, &blob_entry);
+      if (s.ok() && blob_entry.GetValue()) {
+        assert(statuses[i]);
+        assert(blob_entry.GetValue()->size() == value_sizes[i]);
+
+        *statuses[i] = s;
+        blobs[i]->PinSelf(*blob_entry.GetValue());
+
+        // Update the counter for the number of valid blobs read from the cache.
+        ++cached_blob_count;
+        total_bytes += value_sizes[i];
+        cache_hit_mask |= (Mask{1} << i);  // cache hit
+      }
+    }
+
+    // All blobs were read from the cache.
+    if (cached_blob_count == num_blobs) {
+      if (bytes_read) {
+        *bytes_read = total_bytes;
+      }
+      return;
+    }
+  }
+
+  const bool no_io = read_options.read_tier == kBlockCacheTier;
+  if (no_io) {
+    for (size_t i = 0; i < num_blobs; ++i) {
+      if (!(cache_hit_mask & (Mask{1} << i))) {
+        assert(statuses[i]);
+        *statuses[i] =
+            Status::Incomplete("Cannot read blob(s): no disk I/O allowed");
+      }
+    }
+    return;
+  }
+
+  {
+    // Find the rest of blobs from the file since I/O is allowed.
+    autovector<std::reference_wrapper<const Slice>> _user_keys;
+    autovector<uint64_t> _offsets;
+    autovector<uint64_t> _value_sizes;
+    autovector<Status*> _statuses;
+    autovector<PinnableSlice*> _blobs;
+    uint64_t _bytes_read = 0;
+
+    for (size_t i = 0; i < num_blobs; ++i) {
+      if (!(cache_hit_mask & (Mask{1} << i))) {
+        _user_keys.emplace_back(user_keys[i]);
+        _offsets.push_back(offsets[i]);
+        _value_sizes.push_back(value_sizes[i]);
+        _statuses.push_back(statuses[i]);
+        _blobs.push_back(blobs[i]);
+      }
+    }
+
+    CacheHandleGuard<BlobFileReader> blob_file_reader;
+    s = blob_file_cache_->GetBlobFileReader(file_number, &blob_file_reader);
+    if (!s.ok()) {
+      for (size_t i = 0; i < _blobs.size(); ++i) {
+        assert(_statuses[i]);
+        *_statuses[i] = s;
+      }
+      return;
+    }
+
+    assert(blob_file_reader.GetValue());
+
+    blob_file_reader.GetValue()->MultiGetBlob(read_options, _user_keys,
+                                              _offsets, _value_sizes, _statuses,
+                                              _blobs, &_bytes_read);
+
+    if (read_options.fill_cache) {
+      // If filling cache is allowed and a cache is configured, try to put
+      // the blob(s) to the cache.
+      for (size_t i = 0; i < _blobs.size(); ++i) {
+        if (_statuses[i]->ok()) {
+          CachableEntry<std::string> blob_entry;
+          const CacheKey cache_key = base_cache_key.WithOffset(_offsets[i]);
+          const Slice key = cache_key.AsSlice();
+
+          s = PutBlobIntoCache(key, &blob_entry, _blobs[i]);
+          if (!s.ok()) {
+            *_statuses[i] = s;
+          }
+        }
+      }
+    }
+
+    total_bytes += _bytes_read;
+    if (bytes_read) {
+      *bytes_read = total_bytes;
+    }
+
+    RecordTick(statistics_, BLOB_DB_BLOB_FILE_BYTES_READ, _bytes_read);
+  }
+}
+
 bool BlobSource::TEST_BlobInCache(uint64_t file_number, uint64_t file_size,
                                   uint64_t offset) const {
   const CacheKey cache_key = GetCacheKey(file_number, file_size, offset);

db/blob/blob_source.h

@@ -13,6 +13,7 @@
 #include "rocksdb/cache.h"
 #include "rocksdb/rocksdb_namespace.h"
 #include "table/block_based/cachable_entry.h"
+#include "util/autovector.h"
 
 namespace ROCKSDB_NAMESPACE {
 
@@ -42,6 +43,15 @@ class BlobSource {
                  FilePrefetchBuffer* prefetch_buffer, PinnableSlice* value,
                  uint64_t* bytes_read);
 
+  // Offsets must be sorted in ascending order by caller.
+  void MultiGetBlob(
+      const ReadOptions& read_options,
+      const autovector<std::reference_wrapper<const Slice>>& user_keys,
+      uint64_t file_number, uint64_t file_size,
+      const autovector<uint64_t>& offsets,
+      const autovector<uint64_t>& value_sizes, autovector<Status*>& statuses,
+      autovector<PinnableSlice*>& blobs, uint64_t* bytes_read);
+
   inline Status GetBlobFileReader(
       uint64_t blob_file_number,
       CacheHandleGuard<BlobFileReader>* blob_file_reader) {

db/blob/blob_source_test.cc

@@ -117,6 +117,7 @@ TEST_F(BlobSourceTest, GetBlobsFromCache) {
   options.cf_paths.emplace_back(
       test::PerThreadDBPath(env_, "BlobSourceTest_GetBlobsFromCache"), 0);
   options.enable_blob_files = true;
+  options.create_if_missing = true;
 
   LRUCacheOptions co;
   co.capacity = 2048;
@@ -125,7 +126,7 @@ TEST_F(BlobSourceTest, GetBlobsFromCache) {
   options.blob_cache = NewLRUCache(co);
   options.lowest_used_cache_tier = CacheTier::kVolatileTier;
 
-  Reopen(options);
+  DestroyAndReopen(options);
 
   std::string db_id;
   ASSERT_OK(db_->GetDbIdentity(db_id));
@@ -277,10 +278,12 @@ TEST_F(BlobSourceTest, GetBlobsFromCache) {
       ASSERT_FALSE(blob_source.TEST_BlobInCache(blob_file_number, file_size,
                                                 blob_offsets[i]));
 
-      ASSERT_NOK(blob_source.GetBlob(read_options, keys[i], blob_file_number,
-                                     blob_offsets[i], file_size, blob_sizes[i],
-                                     kNoCompression, prefetch_buffer,
-                                     &values[i], &bytes_read));
+      ASSERT_TRUE(blob_source
+                      .GetBlob(read_options, keys[i], blob_file_number,
+                               blob_offsets[i], file_size, blob_sizes[i],
+                               kNoCompression, prefetch_buffer, &values[i],
+                               &bytes_read)
+                      .IsIncomplete());
       ASSERT_TRUE(values[i].empty());
       ASSERT_EQ(bytes_read, 0);
 
@@ -288,6 +291,271 @@ TEST_F(BlobSourceTest, GetBlobsFromCache) {
                                                 blob_offsets[i]));
     }
   }
+
+  {
+    // GetBlob from non-existing file
+    std::vector<PinnableSlice> values(keys.size());
+    uint64_t bytes_read = 0;
+    uint64_t file_number = 100;  // non-existing file
+
+    read_options.read_tier = ReadTier::kReadAllTier;
+    read_options.fill_cache = true;
+
+    for (size_t i = 0; i < num_blobs; ++i) {
+      ASSERT_FALSE(blob_source.TEST_BlobInCache(file_number, file_size,
+                                                blob_offsets[i]));
+
+      ASSERT_TRUE(blob_source
+                      .GetBlob(read_options, keys[i], file_number,
+                               blob_offsets[i], file_size, blob_sizes[i],
+                               kNoCompression, prefetch_buffer, &values[i],
+                               &bytes_read)
+                      .IsIOError());
+      ASSERT_TRUE(values[i].empty());
+      ASSERT_EQ(bytes_read, 0);
+
+      ASSERT_FALSE(blob_source.TEST_BlobInCache(file_number, file_size,
+                                                blob_offsets[i]));
+    }
+  }
+}
+
+TEST_F(BlobSourceTest, MultiGetBlobsFromCache) {
+  Options options;
+  options.env = env_;
+  options.cf_paths.emplace_back(
+      test::PerThreadDBPath(env_, "BlobSourceTest_MultiGetBlobsFromCache"), 0);
+  options.enable_blob_files = true;
+  options.create_if_missing = true;
+
+  LRUCacheOptions co;
+  co.capacity = 2048;
+  co.num_shard_bits = 2;
+  co.metadata_charge_policy = kDontChargeCacheMetadata;
+  options.blob_cache = NewLRUCache(co);
+  options.lowest_used_cache_tier = CacheTier::kVolatileTier;
+
+  DestroyAndReopen(options);
+
+  std::string db_id;
+  ASSERT_OK(db_->GetDbIdentity(db_id));
+
+  std::string db_session_id;
+  ASSERT_OK(db_->GetDbSessionId(db_session_id));
+
+  ImmutableOptions immutable_options(options);
+
+  constexpr uint32_t column_family_id = 1;
+  constexpr bool has_ttl = false;
+  constexpr ExpirationRange expiration_range;
+  constexpr uint64_t blob_file_number = 1;
+  constexpr size_t num_blobs = 16;
+
+  std::vector<std::string> key_strs;
+  std::vector<std::string> blob_strs;
+
+  for (size_t i = 0; i < num_blobs; ++i) {
+    key_strs.push_back("key" + std::to_string(i));
+    blob_strs.push_back("blob" + std::to_string(i));
+  }
+
+  std::vector<Slice> keys;
+  std::vector<Slice> blobs;
+
+  uint64_t file_size = BlobLogHeader::kSize;
+  for (size_t i = 0; i < num_blobs; ++i) {
+    keys.push_back({key_strs[i]});
+    blobs.push_back({blob_strs[i]});
+    file_size += BlobLogRecord::kHeaderSize + keys[i].size() + blobs[i].size();
+  }
+  file_size += BlobLogFooter::kSize;
+
+  std::vector<uint64_t> blob_offsets(keys.size());
+  std::vector<uint64_t> blob_sizes(keys.size());
+
+  WriteBlobFile(immutable_options, column_family_id, has_ttl, expiration_range,
+                expiration_range, blob_file_number, keys, blobs, kNoCompression,
+                blob_offsets, blob_sizes);
+
+  constexpr size_t capacity = 10;
+  std::shared_ptr<Cache> backing_cache =
+      NewLRUCache(capacity);  // Blob file cache
+
+  FileOptions file_options;
+  constexpr HistogramImpl* blob_file_read_hist = nullptr;
+
+  std::unique_ptr<BlobFileCache> blob_file_cache(new BlobFileCache(
+      backing_cache.get(), &immutable_options, &file_options, column_family_id,
+      blob_file_read_hist, nullptr /*IOTracer*/));
+
+  BlobSource blob_source(&immutable_options, db_id, db_session_id,
+                         blob_file_cache.get());
+
+  ReadOptions read_options;
+  read_options.verify_checksums = true;
+
+  constexpr FilePrefetchBuffer* prefetch_buffer = nullptr;
+
+  {
+    // MultiGetBlob
+    uint64_t bytes_read = 0;
+
+    autovector<std::reference_wrapper<const Slice>> key_refs;
+    autovector<uint64_t> offsets;
+    autovector<uint64_t> sizes;
+    std::array<Status, num_blobs> statuses_buf;
+    autovector<Status*> statuses;
+    std::array<PinnableSlice, num_blobs> value_buf;
+    autovector<PinnableSlice*> values;
+
+    for (size_t i = 0; i < num_blobs; i += 2) {  // even index
+      key_refs.emplace_back(std::cref(keys[i]));
+      offsets.push_back(blob_offsets[i]);
+      sizes.push_back(blob_sizes[i]);
+      statuses.push_back(&statuses_buf[i]);
+      values.push_back(&value_buf[i]);
+      ASSERT_FALSE(blob_source.TEST_BlobInCache(blob_file_number, file_size,
+                                                blob_offsets[i]));
+    }
+
+    read_options.fill_cache = true;
+    read_options.read_tier = ReadTier::kReadAllTier;
+
+    // Get half of blobs
+    blob_source.MultiGetBlob(read_options, key_refs, blob_file_number,
+                             file_size, offsets, sizes, statuses, values,
+                             &bytes_read);
+
+    for (size_t i = 0; i < num_blobs; ++i) {
+      if (i % 2 == 0) {
+        ASSERT_OK(statuses_buf[i]);
+        ASSERT_EQ(value_buf[i], blobs[i]);
+        ASSERT_TRUE(blob_source.TEST_BlobInCache(blob_file_number, file_size,
+                                                 blob_offsets[i]));
+      } else {
+        statuses_buf[i].PermitUncheckedError();
+        ASSERT_TRUE(value_buf[i].empty());
+        ASSERT_FALSE(blob_source.TEST_BlobInCache(blob_file_number, file_size,
+                                                  blob_offsets[i]));
+      }
+    }
+
+    // Get the rest of blobs
+    for (size_t i = 1; i < num_blobs; i += 2) {  // odd index
+      ASSERT_FALSE(blob_source.TEST_BlobInCache(blob_file_number, file_size,
+                                                blob_offsets[i]));
+
+      ASSERT_OK(blob_source.GetBlob(read_options, keys[i], blob_file_number,
+                                    blob_offsets[i], file_size, blob_sizes[i],
+                                    kNoCompression, prefetch_buffer,
+                                    &value_buf[i], &bytes_read));
+      ASSERT_EQ(value_buf[i], blobs[i]);
+      ASSERT_EQ(bytes_read,
+                blob_sizes[i] + keys[i].size() + BlobLogRecord::kHeaderSize);
+
+      ASSERT_TRUE(blob_source.TEST_BlobInCache(blob_file_number, file_size,
+                                               blob_offsets[i]));
+    }
+
+    // Cache-only MultiGetBlob
+    read_options.read_tier = ReadTier::kBlockCacheTier;
+
+    key_refs.clear();
+    offsets.clear();
+    sizes.clear();
+    statuses.clear();
+    values.clear();
+    for (size_t i = 0; i < num_blobs; ++i) {
+      key_refs.emplace_back(std::cref(keys[i]));
+      offsets.push_back(blob_offsets[i]);
+      sizes.push_back(blob_sizes[i]);
+      statuses.push_back(&statuses_buf[i]);
+      values.push_back(&value_buf[i]);
+    }
+
+    blob_source.MultiGetBlob(read_options, key_refs, blob_file_number,
+                             file_size, offsets, sizes, statuses, values,
+                             &bytes_read);
+
+    for (size_t i = 0; i < num_blobs; ++i) {
+      ASSERT_OK(statuses_buf[i]);
+      ASSERT_EQ(value_buf[i], blobs[i]);
+      ASSERT_TRUE(blob_source.TEST_BlobInCache(blob_file_number, file_size,
+                                               blob_offsets[i]));
+    }
+  }
+
+  options.blob_cache->EraseUnRefEntries();
+
+  {
+    // Cache-only MultiGetBlob
+    uint64_t bytes_read = 0;
+    read_options.read_tier = ReadTier::kBlockCacheTier;
+
+    autovector<std::reference_wrapper<const Slice>> key_refs;
+    autovector<uint64_t> offsets;
+    autovector<uint64_t> sizes;
+    std::array<Status, num_blobs> statuses_buf;
+    autovector<Status*> statuses;
+    std::array<PinnableSlice, num_blobs> value_buf;
+    autovector<PinnableSlice*> values;
+
+    for (size_t i = 0; i < num_blobs; i++) {
+      key_refs.emplace_back(std::cref(keys[i]));
+      offsets.push_back(blob_offsets[i]);
+      sizes.push_back(blob_sizes[i]);
+      statuses.push_back(&statuses_buf[i]);
+      values.push_back(&value_buf[i]);
+      ASSERT_FALSE(blob_source.TEST_BlobInCache(blob_file_number, file_size,
+                                                blob_offsets[i]));
+    }
+
+    blob_source.MultiGetBlob(read_options, key_refs, blob_file_number,
+                             file_size, offsets, sizes, statuses, values,
+                             &bytes_read);
+
+    for (size_t i = 0; i < num_blobs; ++i) {
+      ASSERT_TRUE(statuses_buf[i].IsIncomplete());
+      ASSERT_TRUE(value_buf[i].empty());
+      ASSERT_FALSE(blob_source.TEST_BlobInCache(blob_file_number, file_size,
+                                                blob_offsets[i]));
+    }
+  }
+
+  {
+    // MultiGetBlob from non-existing file
+    uint64_t bytes_read = 0;
+    uint64_t file_number = 100;  // non-existing file
+    read_options.read_tier = ReadTier::kReadAllTier;
+
+    autovector<std::reference_wrapper<const Slice>> key_refs;
+    autovector<uint64_t> offsets;
+    autovector<uint64_t> sizes;
+    std::array<Status, num_blobs> statuses_buf;
+    autovector<Status*> statuses;
+    std::array<PinnableSlice, num_blobs> value_buf;
+    autovector<PinnableSlice*> values;
+
+    for (size_t i = 0; i < num_blobs; i++) {
+      key_refs.emplace_back(std::cref(keys[i]));
+      offsets.push_back(blob_offsets[i]);
+      sizes.push_back(blob_sizes[i]);
+      statuses.push_back(&statuses_buf[i]);
+      values.push_back(&value_buf[i]);
+      ASSERT_FALSE(blob_source.TEST_BlobInCache(file_number, file_size,
+                                                blob_offsets[i]));
+    }
+
+    blob_source.MultiGetBlob(read_options, key_refs, file_number, file_size,
+                             offsets, sizes, statuses, values, &bytes_read);
+
+    for (size_t i = 0; i < num_blobs; ++i) {
+      ASSERT_TRUE(statuses_buf[i].IsIOError());
+      ASSERT_TRUE(value_buf[i].empty());
+      ASSERT_FALSE(blob_source.TEST_BlobInCache(file_number, file_size,
+                                                blob_offsets[i]));
+    }
+  }
 }
 
 }  // namespace ROCKSDB_NAMESPACE