CBDD/tests/CBDD.Tests.Benchmark/CompressionBenchmarks.cs

using BenchmarkDotNet.Attributes;
using BenchmarkDotNet.Configs;
using BenchmarkDotNet.Jobs;
using System.IO.Compression;
using ZB.MOM.WW.CBDD.Bson;
using ZB.MOM.WW.CBDD.Core.Collections;
using ZB.MOM.WW.CBDD.Core.Compression;
using ZB.MOM.WW.CBDD.Core.Storage;

namespace ZB.MOM.WW.CBDD.Tests.Benchmark;

[SimpleJob]
[InProcess]
[MemoryDiagnoser]
[GroupBenchmarksBy(BenchmarkLogicalGroupRule.ByCategory)]
[HtmlExporter]
[JsonExporterAttribute.Full]
public class CompressionBenchmarks
{
    private const int SeedCount = 300;
    private const int WorkloadCount = 100;

    /// <summary>
    /// Gets or sets whether compression is enabled for the benchmark run.
    /// </summary>
    [Params(false, true)]
    public bool EnableCompression { get; set; }

    /// <summary>
    /// Gets or sets the compression codec for the benchmark run.
    /// </summary>
    [Params(CompressionCodec.Brotli, CompressionCodec.Deflate)]
    public CompressionCodec Codec { get; set; }

    /// <summary>
    /// Gets or sets the compression level for the benchmark run.
    /// </summary>
    [Params(CompressionLevel.Fastest, CompressionLevel.Optimal)]
    public CompressionLevel Level { get; set; }

    private string _dbPath = string.Empty;
    private string _walPath = string.Empty;
    private StorageEngine _storage = null!;
    private BenchmarkTransactionHolder _transactionHolder = null!;
    private DocumentCollection<Person> _collection = null!;

    private Person[] _insertBatch = Array.Empty<Person>();
    private ObjectId[] _seedIds = Array.Empty<ObjectId>();

    /// <summary>
    /// Prepares benchmark storage and seed data for each iteration.
    /// </summary>
    [IterationSetup]
    public void Setup()
    {
        var id = Guid.NewGuid().ToString("N");
        _dbPath = Path.Combine(AppContext.BaseDirectory, $"bench_compression_{id}.db");
        _walPath = Path.ChangeExtension(_dbPath, ".wal");

        var compressionOptions = new CompressionOptions
        {
            EnableCompression = EnableCompression,
            MinSizeBytes = 256,
            MinSavingsPercent = 0,
            Codec = Codec,
            Level = Level
        };

        _storage = new StorageEngine(_dbPath, PageFileConfig.Default, compressionOptions);
        _transactionHolder = new BenchmarkTransactionHolder(_storage);
        _collection = new DocumentCollection<Person>(_storage, _transactionHolder, new PersonMapper());

        _seedIds = new ObjectId[SeedCount];
        for (var i = 0; i < SeedCount; i++)
        {
            var doc = CreatePerson(i, includeLargeBio: true);
            _seedIds[i] = _collection.Insert(doc);
        }

        _transactionHolder.CommitAndReset();

        _insertBatch = Enumerable.Range(SeedCount, WorkloadCount)
            .Select(i => CreatePerson(i, includeLargeBio: true))
            .ToArray();
    }

    /// <summary>
    /// Cleans up benchmark resources for each iteration.
    /// </summary>
    [IterationCleanup]
    public void Cleanup()
    {
        _transactionHolder?.Dispose();
        _storage?.Dispose();

        if (File.Exists(_dbPath)) File.Delete(_dbPath);
        if (File.Exists(_walPath)) File.Delete(_walPath);
    }

    /// <summary>
    /// Benchmarks insert workload performance.
    /// </summary>
    [Benchmark(Baseline = true)]
    [BenchmarkCategory("Compression_InsertUpdateRead")]
    public void Insert_Workload()
    {
        _collection.InsertBulk(_insertBatch);
        _transactionHolder.CommitAndReset();
    }

    /// <summary>
    /// Benchmarks update workload performance.
    /// </summary>
    [Benchmark]
    [BenchmarkCategory("Compression_InsertUpdateRead")]
    public void Update_Workload()
    {
        for (var i = 0; i < WorkloadCount; i++)
        {
            var id = _seedIds[i];
            var current = _collection.FindById(id);
            if (current == null)
                continue;

            current.Bio = BuildBio(i + 10_000);
            current.Age += 1;
            _collection.Update(current);
        }

        _transactionHolder.CommitAndReset();
    }

    /// <summary>
    /// Benchmarks read workload performance.
    /// </summary>
    [Benchmark]
    [BenchmarkCategory("Compression_InsertUpdateRead")]
    public int Read_Workload()
    {
        var checksum = 0;
        for (var i = 0; i < WorkloadCount; i++)
        {
            var person = _collection.FindById(_seedIds[i]);
            if (person != null)
            {
                checksum += person.Age;
            }
        }

        _transactionHolder.CommitAndReset();
        return checksum;
    }

    private static Person CreatePerson(int i, bool includeLargeBio)
    {
        return new Person
        {
            Id = ObjectId.NewObjectId(),
            FirstName = $"First_{i}",
            LastName = $"Last_{i}",
            Age = 20 + (i % 50),
            Bio = includeLargeBio ? BuildBio(i) : $"bio-{i}",
            CreatedAt = DateTime.UnixEpoch.AddMinutes(i),
            Balance = 100 + i,
            HomeAddress = new Address
            {
                Street = $"{i} Main St",
                City = "Bench City",
                ZipCode = "12345"
            },
            EmploymentHistory =
            [
                new WorkHistory
                {
                    CompanyName = $"Company_{i}",
                    Title = "Engineer",
                    DurationYears = i % 10,
                    Tags = ["csharp", "db", "compression"]
                }
            ]
        };
    }

    private static string BuildBio(int seed)
    {
        var builder = new System.Text.StringBuilder(4500);
        for (var i = 0; i < 150; i++)
        {
            builder.Append("bio-");
            builder.Append(seed.ToString("D6"));
            builder.Append('-');
            builder.Append(i.ToString("D3"));
            builder.Append('|');
        }

        return builder.ToString();
    }
}