feat(storage): faster InsertObject() uploads

google/cloud/storage/internal/curl_client.cc

@@ -1304,16 +1304,13 @@ StatusOr<ObjectMetadata> CurlClient::InsertObjectMediaMultipart(
 }
 
 std::string CurlClient::PickBoundary(std::string const& text_to_avoid) {
-  // We need to find a string that is *not* found in `text_to_avoid`, we pick
-  // a string at random, and see if it is in `text_to_avoid`, if it is, we grow
-  // the string with random characters and start from where we last found a
-  // the candidate.  Eventually we will find something, though it might be
-  // larger than `text_to_avoid`.  And we only make (approximately) one pass
-  // over `text_to_avoid`.
   auto generate_candidate = [this]() {
     std::unique_lock<std::mutex> lk(mu_);
     return GenerateMessageBoundaryCandidate(generator_);
   };
+  if (!CurrentOptions().get<ValidateInsertObjectBoundary>()) {
+    return generate_candidate();
+  }
   return GenerateMessageBoundary(text_to_avoid, generate_candidate);
 }
 

google/cloud/storage/internal/generate_message_boundary.cc

@@ -33,10 +33,11 @@ std::string GenerateMessageBoundary(
 
 std::string GenerateMessageBoundaryCandidate(
     google::cloud::internal::DefaultPRNG& generator) {
-  auto candidate = std::string{
-      "abcdefghijklmnopqrstuvwxyz0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ"};
-  std::shuffle(candidate.begin(), candidate.end(), generator);
-  return candidate;
+  auto constexpr kCandidateLength = 64;
+  return google::cloud::internal::Sample(generator, kCandidateLength,
+                                         "ABCDEFGHIJKLMNOPQRSTUVWXYZ"
+                                         "abcdefghijklmnopqrstuvwxyz"
+                                         "0123456789");
 }
 
 }  // namespace internal

google/cloud/storage/internal/generate_message_boundary_benchmark.cc

@@ -37,29 +37,15 @@ namespace {
 //  L1 Instruction 32 KiB (x48)
 //  L2 Unified 1024 KiB (x48)
 //  L3 Unified 39424 KiB (x2)
-// ---------------------------------------------------------------------------------------------
-// Benchmark                                                   Time             CPU   Iterations
-// ---------------------------------------------------------------------------------------------
-// ...
-// GenerateBoundaryFixture/GenerateBoundary_mean        20478679 ns     20475235 ns          100
-// GenerateBoundaryFixture/GenerateBoundary_median      20583724 ns     20580943 ns          100
-// GenerateBoundaryFixture/GenerateBoundary_stddev        335315 ns       335459 ns          100
-// GenerateBoundaryFixture/GenerateBoundary_cv              1.64 %          1.64 %           100
-// ...
-// GenerateBoundaryFixture/GenerateBoundaryOld_mean     20809894 ns     20806317 ns          100
-// GenerateBoundaryFixture/GenerateBoundaryOld_median   20520133 ns     20517279 ns          100
-// GenerateBoundaryFixture/GenerateBoundaryOld_stddev    1284277 ns      1284334 ns          100
-// GenerateBoundaryFixture/GenerateBoundaryOld_cv           6.17 %          6.17 %           100
-// ...
-// GenerateBoundaryFixture/WorstCase_mean              100747489 ns    100727911 ns          100
-// GenerateBoundaryFixture/WorstCase_median            101026913 ns    101006689 ns          100
-// GenerateBoundaryFixture/WorstCase_stddev              1285934 ns      1285884 ns          100
-// GenerateBoundaryFixture/WorstCase_cv                     1.28 %          1.28 %           100
-// ...
-// GenerateBoundaryFixture/BestCase_mean                 9584895 ns      9583080 ns          100
-// GenerateBoundaryFixture/BestCase_median               9598452 ns      9597243 ns          100
-// GenerateBoundaryFixture/BestCase_stddev                 90679 ns        90643 ns          100
-// GenerateBoundaryFixture/BestCase_cv                      0.95 %          0.95 %           100
+// Load Average: 8.44, 25.15, 23.46
+// -------------------------------------------------------------------------------------------------
+// Benchmark                                                       Time             CPU   Iterations
+// -------------------------------------------------------------------------------------------------
+// GenerateBoundaryFixture/GenerateBoundary                      505 ns          505 ns      1385317
+// GenerateBoundaryFixture/GenerateBoundaryWithValidation   20031391 ns     20025303 ns           35
+// GenerateBoundaryFixture/GenerateBoundaryOld              20133230 ns     20129379 ns           35
+// GenerateBoundaryFixture/WorstCase                       100998844 ns    100985746 ns            7
+// GenerateBoundaryFixture/BestCase                          9739599 ns      9736802 ns           69
 // clang-format on
 
 auto constexpr kMessageSize = 128 * 1024 * 1024;
@@ -100,6 +86,15 @@ BENCHMARK_F(GenerateBoundaryFixture, GenerateBoundary)
 (benchmark::State& state) {
   auto make_string = [this]() { return GenerateCandidate(); };
 
+  for (auto _ : state) {
+    benchmark::DoNotOptimize(make_string());
+  }
+}
+
+BENCHMARK_F(GenerateBoundaryFixture, GenerateBoundaryWithValidation)
+(benchmark::State& state) {
+  auto make_string = [this]() { return GenerateCandidate(); };
+
   for (auto _ : state) {
     benchmark::DoNotOptimize(GenerateMessageBoundary(message(), make_string));
   }

google/cloud/storage/internal/rest_client.cc

@@ -338,6 +338,9 @@ std::string RestClient::PickBoundary(std::string const& text_to_avoid) {
     std::unique_lock<std::mutex> lk(mu_);
     return GenerateMessageBoundaryCandidate(generator_);
   };
+  if (!CurrentOptions().get<ValidateInsertObjectBoundary>()) {
+    return generate_candidate();
+  }
   return GenerateMessageBoundary(text_to_avoid, generate_candidate);
 }
 

google/cloud/storage/options.h

@@ -286,6 +286,30 @@ struct IdempotencyPolicyOption {
   using Type = std::shared_ptr<IdempotencyPolicy>;
 };
 
+/**
+ * Prevent
+ *
+ * The `InsertObject()` function often uses JSON multipart uploads, i.e, a
+ * type of [single request upload] where the HTTP payload contains both the
+ * metadata and data for the object. These uploads require a message boundary
+ * that is not found in the payload. By default, the client library will
+ * generate a random separator and optimistically assume it will not be found in
+ * the payload. In the worst case scenario, an invalid separator would be
+ * detected by the service, and result in an error.
+ *
+ * When this option is enabled the client library will validate the message,
+ * if it finds a collision, it will repeatedly generate new random strings until
+ * one that does not collide is found. The space of possible random strings is
+ * large enough this should be found in one or two attempts.
+ *
+ * [single request upload]:
+ * https://cloud.google.com/storage/docs/uploads-downloads [JSON multipart
+ * uploads]: https://cloud.google.com/storage/docs/json_api/v1/objects/insert
+ */
+struct ValidateInsertObjectBoundary {
+  using Type = bool;
+};
+
 /// The complete list of options accepted by `storage::Client`.
 using ClientOptionList = ::google::cloud::OptionList<
     RestEndpointOption, IamEndpointOption, Oauth2CredentialsOption,
@@ -295,7 +319,7 @@ using ClientOptionList = ::google::cloud::OptionList<
     MaximumCurlSocketRecvSizeOption, MaximumCurlSocketSendSizeOption,
     TransferStallTimeoutOption, RetryPolicyOption, BackoffPolicyOption,
     IdempotencyPolicyOption, CARootsFilePathOption,
-    storage_experimental::HttpVersionOption>;
+    ValidateInsertObjectBoundary, storage_experimental::HttpVersionOption>;
 
 GOOGLE_CLOUD_CPP_INLINE_NAMESPACE_END
 }  // namespace storage