Compare fog vs storage-cli benchmarks (#4738)

* Compare fog vs storage-cli benchmarks

Author: kathap

lib/cloud_controller/benchmark/blobstore.rb

@@ -1,48 +1,70 @@
+# frozen_string_literal: true
+
 require 'benchmark'
 require 'find'
 require 'zip'
+require 'tempfile'
+require 'fileutils'
+require 'securerandom'
 
 module VCAP::CloudController
   module Benchmark
     class Blobstore
+      SIZES = {
+        '0.005MB' => (0.005 * 1024 * 1024).to_i,
+        '0.01MB' => (0.01 * 1024 * 1024).to_i,
+        '0.1MB' => (0.1 * 1024 * 1024).to_i,
+        '1MB' => 1 * 1024 * 1024,
+        '10MB' => 10   * 1024 * 1024,
+        '50MB' => 50   * 1024 * 1024,
+        '100MB' => 100  * 1024 * 1024,
+        '500MB' => 400  * 1024 * 1024,
+        '1000MB' => 1000 * 1024 * 1024
+      }.freeze
+
+      CHUNK_1MB = '0'.b * (1024 * 1024)
+
       def perform
+        big_droplet_guids = []
         resource_dir = generate_resources
-
-        resource_timing = resource_match(resource_dir)
-        puts("resource match timing: #{resource_timing * 1000}ms")
+        log_timing('resource match timing', resource_match(resource_dir))
 
         zip_output_dir = Dir.mktmpdir
         zip_file = zip_resources(resource_dir, zip_output_dir)
 
-        package_guid, resource_timing = upload_package(zip_file)
-        puts("package upload timing: #{resource_timing * 1000}ms")
-
-        resource_timing = download_package(package_guid, resource_dir)
-        puts("package download timing: #{resource_timing * 1000}ms")
+        package_guid, timing = upload_package(zip_file)
+        log_timing('package upload timing', timing)
+        log_timing('package download timing', download_package(package_guid, resource_dir))
 
-        bytes_read, resource_timing = download_buildpacks(resource_dir)
+        bytes_read, timing = download_buildpacks(resource_dir)
         puts("downloaded #{Buildpack.count} buildpacks, total #{bytes_read} bytes read")
-        puts("buildpack download timing: #{resource_timing * 1000}ms")
+        log_timing('buildpack download timing', timing)
 
-        droplet_guid, resource_timing = upload_droplet(zip_file)
-        puts("droplet upload timing: #{resource_timing * 1000}ms")
+        upload_lines = []
+        download_lines = []
 
-        resource_timing = download_droplet(droplet_guid, resource_dir)
-        puts("droplet download timing: #{resource_timing * 1000}ms")
+        SIZES.each do |label, bytes|
+          Tempfile.create(["big-droplet-#{label}", '.bin'], resource_dir) do |tempfile|
+            write_file_of_size(tempfile.path, bytes)
 
-        big_droplet_file = Tempfile.new('big-droplet', resource_dir)
-        big_droplet_file.write('abc' * 1024 * 1024 * 100)
-        big_droplet_guid, resource_timing = upload_droplet(big_droplet_file.path)
-        puts("big droplet upload timing: #{resource_timing * 1000}ms")
+            guid, upload_timing = upload_droplet(tempfile.path)
+            big_droplet_guids << guid
+
+            download_timing = download_droplet(guid, resource_dir)
+
+            upload_lines << format_timing("droplet #{label} upload timing", upload_timing)
+            download_lines << format_timing("droplet #{label} download timing", download_timing)
+          end
+        end
 
-        resource_timing = download_droplet(big_droplet_guid, resource_dir)
-        puts("big droplet download timing: #{resource_timing * 1000}ms")
+        puts(upload_lines.join("\n"))
+        puts(download_lines.join("\n"))
       ensure
         FileUtils.remove_dir(resource_dir, true)
         FileUtils.remove_dir(zip_output_dir, true)
-        package_blobstore_client.delete(package_guid) if package_guid
-        droplet_blobstore_client.delete(droplet_guid) if droplet_guid
-        droplet_blobstore_client.delete(big_droplet_guid) if big_droplet_guid
+
+        safe_delete(package_blobstore_client, package_guid)
+        Array(big_droplet_guids).each { |g| safe_delete(droplet_blobstore_client, g) }
       end
 
       def resource_match(dir_path)
@@ -60,46 +82,73 @@ def upload_package(package_path)
       end
 
       def download_package(package_guid, tmp_dir)
-        tempfile = Tempfile.new('package-download-benchmark', tmp_dir)
-        ::Benchmark.realtime do
-          package_blobstore_client.download_from_blobstore(package_guid, tempfile.path)
+        Tempfile.create('package-download-benchmark', tmp_dir) do |tempfile|
+          ::Benchmark.realtime do
+            package_blobstore_client.download_from_blobstore(package_guid, tempfile.path)
+          end
         end
       end
 
       def download_buildpacks(tmp_dir)
-        tempfile = Tempfile.new('buildpack-download-benchmark', tmp_dir)
-        bytes_read = 0
-
-        timing = ::Benchmark.realtime do
-          bytes_read = Buildpack.map do |buildpack|
-            buildpack_blobstore_client.download_from_blobstore(buildpack.key, tempfile.path)
-            File.stat(tempfile.path).size
-          end.sum
+        Tempfile.create('buildpack-download-benchmark', tmp_dir) do |tempfile|
+          bytes_read = 0
+          timing = ::Benchmark.realtime do
+            bytes_read = Buildpack.map do |buildpack|
+              buildpack_blobstore_client.download_from_blobstore(buildpack.key, tempfile.path)
+              File.stat(tempfile.path).size
+            end.sum
+          end
+          [bytes_read, timing]
         end
-
-        [bytes_read, timing]
       end
 
       def upload_droplet(droplet_path)
         copy_to_blobstore(droplet_path, droplet_blobstore_client)
       end
 
       def download_droplet(droplet_guid, tmp_dir)
-        tempfile = Tempfile.new('droplet-download-benchmark', tmp_dir)
-
-        ::Benchmark.realtime do
-          droplet_blobstore_client.download_from_blobstore(droplet_guid, tempfile.path)
+        Tempfile.create('droplet-download-benchmark', tmp_dir) do |tempfile|
+          ::Benchmark.realtime do
+            droplet_blobstore_client.download_from_blobstore(droplet_guid, tempfile.path)
+          end
         end
       end
 
       private
 
+      def log_timing(label, seconds)
+        puts("#{label}: #{(seconds * 1000).round(3)}ms")
+      end
+
+      def format_timing(label, seconds)
+        "#{label}: #{(seconds * 1000).round(3)}ms"
+      end
+
+      def safe_delete(client, guid)
+        return if guid.nil?
+
+        client.delete(guid)
+      rescue StandardError => e
+        # don't fail the benchmark run if cleanup fails
+        warn("cleanup failed for guid=#{guid}: #{e.class}: #{e.message}")
+      end
+
+      def write_file_of_size(path, bytes)
+        File.open(path, 'wb') do |f|
+          remaining = bytes
+          while remaining > 0
+            to_write = [CHUNK_1MB.bytesize, remaining].min
+            f.write(CHUNK_1MB, to_write)
+            remaining -= to_write
+          end
+        end
+      end
+
       def generate_resources
         dir = Dir.mktmpdir
 
-        100.times.each do |i|
-          f = File.open(File.join(dir, i.to_s), 'w')
-          f.write('foo' * (65_536 + i))
+        100.times do |i|
+          File.write(File.join(dir, i.to_s), 'foo' * (65_536 + i))
         end
 
         dir
@@ -110,9 +159,7 @@ def zip_resources(resource_dir, output_dir)
         Zip::File.open(zip_file, create: true) do |zipfile|
           Find.find(resource_dir).
             select { |f| File.file?(f) }.
-            each do |file|
-            zipfile.add(File.basename(file), file)
-          end
+            each { |file| zipfile.add(File.basename(file), file) }
         end
         zip_file
       end

lib/cloud_controller/config.rb

@@ -12,6 +12,7 @@
 require 'cloud_controller/config_schemas/worker_schema'
 require 'cloud_controller/config_schemas/deployment_updater_schema'
 require 'cloud_controller/config_schemas/rotate_database_key_schema'
+require 'cloud_controller/config_schemas/blobstore_benchmarks_schema'
 require 'utils/hash_utils'
 
 module VCAP::CloudController

lib/cloud_controller/config_schemas/blobstore_benchmarks_schema.rb

@@ -0,0 +1,81 @@
+require 'vcap/config'
+
+module VCAP::CloudController
+  module ConfigSchemas
+    class BlobstoreBenchmarksSchema < VCAP::Config
+      # rubocop:disable Metrics/BlockLength
+      define_schema do
+        blobstore_section = {
+          blobstore_type: String,
+          blobstore_provider: String,
+
+          optional(:connection_config) => Hash,
+          optional(:fog_connection) => Hash,
+
+          fog_aws_storage_options: Hash,
+          fog_gcp_storage_options: Hash,
+
+          optional(:resource_directory_key) => String,
+          optional(:buildpack_directory_key) => String,
+          optional(:app_package_directory_key) => String,
+          optional(:droplet_directory_key) => String,
+
+          optional(:maximum_size) => Integer,
+          optional(:minimum_size) => Integer,
+          optional(:max_package_size) => Integer,
+          optional(:max_valid_packages_stored) => Integer,
+          optional(:max_staged_droplets_stored) => Integer
+        }
+
+        {
+          optional(:logging) => {
+            optional(:level) => String,
+            optional(:file) => String,
+            optional(:syslog) => String,
+            optional(:stdout_sink_enabled) => bool
+          },
+
+          db: {
+            optional(:database) => Hash, # db connection hash for sequel\
+            max_connections: Integer, # max connections in the connection pool
+            pool_timeout: Integer, # timeout before raising an error when connection can't be established to the db
+            log_level: String, # debug, info, etc.
+            log_db_queries: bool,
+            ssl_verify_hostname: bool,
+            connection_validation_timeout: Integer,
+            optional(:ca_cert_path) => String
+          },
+          storage_cli_config_file_resource_pool: String,
+          storage_cli_config_file_buildpacks: String,
+          storage_cli_config_file_packages: String,
+          storage_cli_config_file_droplets: String,
+
+          db_encryption_key: enum(String, NilClass),
+
+          optional(:database_encryption) => {
+            keys: Hash,
+            current_key_label: String,
+            optional(:pbkdf2_hmac_iterations) => Integer
+          },
+
+          resource_pool: blobstore_section,
+          buildpacks: blobstore_section,
+          packages: blobstore_section,
+          droplets: blobstore_section,
+
+          pid_filename: String,
+          index: Integer, # Component index (cc-0, cc-1, etc)
+          name: String, # Component name (api_z1, api_z2)
+          default_app_ssh_access: bool
+        }
+      end
+      # rubocop:enable Metrics/BlockLength
+
+      class << self
+        def configure_components(config)
+          ResourcePool.instance = ResourcePool.new(config)
+        end
+      end
+    end
+  end
+end

lib/tasks/blobstore_benchmarks.rake

@@ -3,7 +3,8 @@ require 'cloud_controller/benchmark/blobstore'
 namespace :benchmarks do
   desc 'Perform blobstore benchmark'
   task perform_blobstore_benchmark: :environment do
-    BackgroundJobEnvironment.new(RakeConfig.config).setup_environment do
+    RakeConfig.context = :blobstore_benchmarks
+    BoshErrandEnvironment.new(RakeConfig.config).setup_environment do
       VCAP::CloudController::Benchmark::Blobstore.new.perform
     end
   end