Fastest way to serialize and deserialize .NET objects

A comprehensive comparison between different formats made by me in this post- https://medium.com/@maximn/serialization-performance-comparison-xml-binary-json-p-ad737545d227

Just one sample from the post-

Having an interest in this, I decided to test the suggested methods with the closest "apples to apples" test I could. I wrote a Console app, with the following code:

using System; using System.Collections.Generic; using System.Diagnostics; using System.IO; using System.Linq; using System.Runtime.Serialization.Formatters.Binary; using System.Text; using System.Threading.Tasks; namespace SerializationTests { class Program { static void Main(string[] args) { var count = 100000; var rnd = new Random(DateTime.UtcNow.GetHashCode()); Console.WriteLine("Generating {0} arrays of data...", count); var arrays = new List<int[]>(); for (int i = 0; i < count; i++) { var elements = rnd.Next(1, 100); var array = new int[elements]; for (int j = 0; j < elements; j++) { array[j] = rnd.Next(); } arrays.Add(array); } Console.WriteLine("Test data generated."); var stopWatch = new Stopwatch(); Console.WriteLine("Testing BinarySerializer..."); var binarySerializer = new BinarySerializer(); var binarySerialized = new List<byte[]>(); var binaryDeserialized = new List<int[]>(); stopWatch.Reset(); stopWatch.Start(); foreach (var array in arrays) { binarySerialized.Add(binarySerializer.Serialize(array)); } stopWatch.Stop(); Console.WriteLine("BinaryFormatter: Serializing took {0}ms.", stopWatch.Elapsed.TotalMilliseconds); stopWatch.Reset(); stopWatch.Start(); foreach (var serialized in binarySerialized) { binaryDeserialized.Add(binarySerializer.Deserialize<int[]>(serialized)); } stopWatch.Stop(); Console.WriteLine("BinaryFormatter: Deserializing took {0}ms.", stopWatch.Elapsed.TotalMilliseconds); Console.WriteLine(); Console.WriteLine("Testing ProtoBuf serializer..."); var protobufSerializer = new ProtoBufSerializer(); var protobufSerialized = new List<byte[]>(); var protobufDeserialized = new List<int[]>(); stopWatch.Reset(); stopWatch.Start(); foreach (var array in arrays) { protobufSerialized.Add(protobufSerializer.Serialize(array)); } stopWatch.Stop(); Console.WriteLine("ProtoBuf: Serializing took {0}ms.", stopWatch.Elapsed.TotalMilliseconds); stopWatch.Reset(); stopWatch.Start(); foreach (var serialized in protobufSerialized) { protobufDeserialized.Add(protobufSerializer.Deserialize<int[]>(serialized)); } stopWatch.Stop(); Console.WriteLine("ProtoBuf: Deserializing took {0}ms.", stopWatch.Elapsed.TotalMilliseconds); Console.WriteLine(); Console.WriteLine("Testing NetSerializer serializer..."); var netSerializerSerializer = new ProtoBufSerializer(); var netSerializerSerialized = new List<byte[]>(); var netSerializerDeserialized = new List<int[]>(); stopWatch.Reset(); stopWatch.Start(); foreach (var array in arrays) { netSerializerSerialized.Add(netSerializerSerializer.Serialize(array)); } stopWatch.Stop(); Console.WriteLine("NetSerializer: Serializing took {0}ms.", stopWatch.Elapsed.TotalMilliseconds); stopWatch.Reset(); stopWatch.Start(); foreach (var serialized in netSerializerSerialized) { netSerializerDeserialized.Add(netSerializerSerializer.Deserialize<int[]>(serialized)); } stopWatch.Stop(); Console.WriteLine("NetSerializer: Deserializing took {0}ms.", stopWatch.Elapsed.TotalMilliseconds); Console.WriteLine("Press any key to end."); Console.ReadKey(); } public class BinarySerializer { private static readonly BinaryFormatter Formatter = new BinaryFormatter(); public byte[] Serialize(object toSerialize) { using (var stream = new MemoryStream()) { Formatter.Serialize(stream, toSerialize); return stream.ToArray(); } } public T Deserialize<T>(byte[] serialized) { using (var stream = new MemoryStream(serialized)) { var result = (T)Formatter.Deserialize(stream); return result; } } } public class ProtoBufSerializer { public byte[] Serialize(object toSerialize) { using (var stream = new MemoryStream()) { ProtoBuf.Serializer.Serialize(stream, toSerialize); return stream.ToArray(); } } public T Deserialize<T>(byte[] serialized) { using (var stream = new MemoryStream(serialized)) { var result = ProtoBuf.Serializer.Deserialize<T>(stream); return result; } } } public class NetSerializer { private static readonly NetSerializer Serializer = new NetSerializer(); public byte[] Serialize(object toSerialize) { return Serializer.Serialize(toSerialize); } public T Deserialize<T>(byte[] serialized) { return Serializer.Deserialize<T>(serialized); } } } } 

The results surprised me; they were consistent when run multiple times:

Generating 100000 arrays of data... Test data generated. Testing BinarySerializer... BinaryFormatter: Serializing took 336.8392ms. BinaryFormatter: Deserializing took 208.7527ms. Testing ProtoBuf serializer... ProtoBuf: Serializing took 2284.3827ms. ProtoBuf: Deserializing took 2201.8072ms. Testing NetSerializer serializer... NetSerializer: Serializing took 2139.5424ms. NetSerializer: Deserializing took 2113.7296ms. Press any key to end. 

Collecting these results, I decided to see if ProtoBuf or NetSerializer performed better with larger objects. I changed the collection count to 10,000 objects, but increased the size of the arrays to 1-10,000 instead of 1-100. The results seemed even more definitive:

Generating 10000 arrays of data... Test data generated. Testing BinarySerializer... BinaryFormatter: Serializing took 285.8356ms. BinaryFormatter: Deserializing took 206.0906ms. Testing ProtoBuf serializer... ProtoBuf: Serializing took 10693.3848ms. ProtoBuf: Deserializing took 5988.5993ms. Testing NetSerializer serializer... NetSerializer: Serializing took 9017.5785ms. NetSerializer: Deserializing took 5978.7203ms. Press any key to end. 

My conclusion, therefore, is: there may be cases where ProtoBuf and NetSerializer are well-suited to, but in terms of raw performance for at least relatively simple objects... BinaryFormatter is significantly more performant, by at least an order of magnitude.

YMMV.

Yet another serializer out there that claims to be super fast is netserializer.

The data given on their site shows performance of 2x - 4x over protobuf, I have not tried this myself, but if you are evaluating various options, try this as well

Protobuf is very very fast.

See http://code.google.com/p/protobuf-net/wiki/Performance for in depth information concerning the performance of this system, and an implementation.

The binary serializer included with .net should be faster that the XmlSerializer. Or another serializer for protobuf, json, ...

But for some of them you need to add Attributes, or some other way to add metadata. For example ProtoBuf uses numeric property IDs internally, and the mapping needs to be somehow conserved by a different mechanism. Versioning isn't trivial with any serializer.

I removed the bugs in above code and got below results: Also I am unsure given how NetSerializer requires you to register the types you are serializing, what kind of compatibility or performance differences that could potentially make.

Generating 100000 arrays of data... Test data generated. Testing BinarySerializer... BinaryFormatter: Serializing took 508.9773ms. BinaryFormatter: Deserializing took 371.8499ms. Testing ProtoBuf serializer... ProtoBuf: Serializing took 3280.9185ms. ProtoBuf: Deserializing took 3190.7899ms. Testing NetSerializer serializer... NetSerializer: Serializing took 427.1241ms. NetSerializer: Deserializing took 78.954ms. Press any key to end. 

Modified Code

using System; using System.Collections.Generic; using System.Diagnostics; using System.IO; using System.Linq; using System.Runtime.Serialization.Formatters.Binary; using System.Text; using System.Threading.Tasks; namespace SerializationTests { class Program { static void Main(string[] args) { var count = 100000; var rnd = new Random((int)DateTime.UtcNow.Ticks & 0xFF); Console.WriteLine("Generating {0} arrays of data...", count); var arrays = new List<int[]>(); for (int i = 0; i < count; i++) { var elements = rnd.Next(1, 100); var array = new int[elements]; for (int j = 0; j < elements; j++) { array[j] = rnd.Next(); } arrays.Add(array); } Console.WriteLine("Test data generated."); var stopWatch = new Stopwatch(); Console.WriteLine("Testing BinarySerializer..."); var binarySerializer = new BinarySerializer(); var binarySerialized = new List<byte[]>(); var binaryDeserialized = new List<int[]>(); stopWatch.Reset(); stopWatch.Start(); foreach (var array in arrays) { binarySerialized.Add(binarySerializer.Serialize(array)); } stopWatch.Stop(); Console.WriteLine("BinaryFormatter: Serializing took {0}ms.", stopWatch.Elapsed.TotalMilliseconds); stopWatch.Reset(); stopWatch.Start(); foreach (var serialized in binarySerialized) { binaryDeserialized.Add(binarySerializer.Deserialize<int[]>(serialized)); } stopWatch.Stop(); Console.WriteLine("BinaryFormatter: Deserializing took {0}ms.", stopWatch.Elapsed.TotalMilliseconds); Console.WriteLine(); Console.WriteLine("Testing ProtoBuf serializer..."); var protobufSerializer = new ProtoBufSerializer(); var protobufSerialized = new List<byte[]>(); var protobufDeserialized = new List<int[]>(); stopWatch.Reset(); stopWatch.Start(); foreach (var array in arrays) { protobufSerialized.Add(protobufSerializer.Serialize(array)); } stopWatch.Stop(); Console.WriteLine("ProtoBuf: Serializing took {0}ms.", stopWatch.Elapsed.TotalMilliseconds); stopWatch.Reset(); stopWatch.Start(); foreach (var serialized in protobufSerialized) { protobufDeserialized.Add(protobufSerializer.Deserialize<int[]>(serialized)); } stopWatch.Stop(); Console.WriteLine("ProtoBuf: Deserializing took {0}ms.", stopWatch.Elapsed.TotalMilliseconds); Console.WriteLine(); Console.WriteLine("Testing NetSerializer serializer..."); var netSerializerSerialized = new List<byte[]>(); var netSerializerDeserialized = new List<int[]>(); stopWatch.Reset(); stopWatch.Start(); var netSerializerSerializer = new NS(); foreach (var array in arrays) { netSerializerSerialized.Add(netSerializerSerializer.Serialize(array)); } stopWatch.Stop(); Console.WriteLine("NetSerializer: Serializing took {0}ms.", stopWatch.Elapsed.TotalMilliseconds); stopWatch.Reset(); stopWatch.Start(); foreach (var serialized in netSerializerSerialized) { netSerializerDeserialized.Add(netSerializerSerializer.Deserialize<int[]>(serialized)); } stopWatch.Stop(); Console.WriteLine("NetSerializer: Deserializing took {0}ms.", stopWatch.Elapsed.TotalMilliseconds); Console.WriteLine("Press any key to end."); Console.ReadKey(); } public class BinarySerializer { private static readonly BinaryFormatter Formatter = new BinaryFormatter(); public byte[] Serialize(object toSerialize) { using (var stream = new MemoryStream()) { Formatter.Serialize(stream, toSerialize); return stream.ToArray(); } } public T Deserialize<T>(byte[] serialized) { using (var stream = new MemoryStream(serialized)) { var result = (T)Formatter.Deserialize(stream); return result; } } } public class ProtoBufSerializer { public byte[] Serialize(object toSerialize) { using (var stream = new MemoryStream()) { ProtoBuf.Serializer.Serialize(stream, toSerialize); return stream.ToArray(); } } public T Deserialize<T>(byte[] serialized) { using (var stream = new MemoryStream(serialized)) { var result = ProtoBuf.Serializer.Deserialize<T>(stream); return result; } } } public class NS { NetSerializer.Serializer Serializer = new NetSerializer.Serializer(new Type[] { typeof(int), typeof(int[]) }); public byte[] Serialize(object toSerialize) { using (var stream = new MemoryStream()) { Serializer.Serialize(stream, toSerialize); return stream.ToArray(); } } public T Deserialize<T>(byte[] serialized) { using (var stream = new MemoryStream(serialized)) { Serializer.Deserialize(stream, out var result); return (T)result; } } } } } 

You can try Salar.Bois serializer which has a decent performance. Its focus is on payload size but it also offers good performance.

There are benchmarks in the Github page if you wish to see and compare the results by yourself.

https://github.com/salarcode/Bois

I took the liberty of feeding your classes into the CGbR generator. Because it is in an early stage it doesn't support DateTime yet, so I simply replaced it with long. The generated serialization code looks like this:

public int Size { get { var size = 24; // Add size for collections and strings size += Cts == null ? 0 : Cts.Count * 4; size += Tes == null ? 0 : Tes.Count * 4; size += Code == null ? 0 : Code.Length; size += Message == null ? 0 : Message.Length; return size; } } public byte[] ToBytes(byte[] bytes, ref int index) { if (index + Size > bytes.Length) throw new ArgumentOutOfRangeException("index", "Object does not fit in array"); // Convert Cts // Two bytes length information for each dimension GeneratorByteConverter.Include((ushort)(Cts == null ? 0 : Cts.Count), bytes, ref index); if (Cts != null) { for(var i = 0; i < Cts.Count; i++) { var value = Cts[i]; value.ToBytes(bytes, ref index); } } // Convert Tes // Two bytes length information for each dimension GeneratorByteConverter.Include((ushort)(Tes == null ? 0 : Tes.Count), bytes, ref index); if (Tes != null) { for(var i = 0; i < Tes.Count; i++) { var value = Tes[i]; value.ToBytes(bytes, ref index); } } // Convert Code GeneratorByteConverter.Include(Code, bytes, ref index); // Convert Message GeneratorByteConverter.Include(Message, bytes, ref index); // Convert StartDate GeneratorByteConverter.Include(StartDate.ToBinary(), bytes, ref index); // Convert EndDate GeneratorByteConverter.Include(EndDate.ToBinary(), bytes, ref index); return bytes; } public Td FromBytes(byte[] bytes, ref int index) { // Read Cts var ctsLength = GeneratorByteConverter.ToUInt16(bytes, ref index); var tempCts = new List<Ct>(ctsLength); for (var i = 0; i < ctsLength; i++) { var value = new Ct().FromBytes(bytes, ref index); tempCts.Add(value); } Cts = tempCts; // Read Tes var tesLength = GeneratorByteConverter.ToUInt16(bytes, ref index); var tempTes = new List<Te>(tesLength); for (var i = 0; i < tesLength; i++) { var value = new Te().FromBytes(bytes, ref index); tempTes.Add(value); } Tes = tempTes; // Read Code Code = GeneratorByteConverter.GetString(bytes, ref index); // Read Message Message = GeneratorByteConverter.GetString(bytes, ref index); // Read StartDate StartDate = DateTime.FromBinary(GeneratorByteConverter.ToInt64(bytes, ref index)); // Read EndDate EndDate = DateTime.FromBinary(GeneratorByteConverter.ToInt64(bytes, ref index)); return this; } 

I created a list of sample objects like this:

var objects = new List<Td>(); for (int i = 0; i < 1000; i++) { var obj = new Td { Message = "Hello my friend", Code = "Some code that can be put here", StartDate = DateTime.Now.AddDays(-7), EndDate = DateTime.Now.AddDays(2), Cts = new List<Ct>(), Tes = new List<Te>() }; for (int j = 0; j < 10; j++) { obj.Cts.Add(new Ct { Foo = i * j }); obj.Tes.Add(new Te { Bar = i + j }); } objects.Add(obj); } 

Results on my machine in Release build:

var watch = new Stopwatch(); watch.Start(); var bytes = BinarySerializer.SerializeMany(objects); watch.Stop(); 

Size: 149000 bytes

Time: 2.059ms 3.13ms

Edit: Starting with CGbR 0.4.3 the binary serializer supports DateTime. Unfortunately the DateTime.ToBinary method is insanely slow. I will replace it with somehting faster soon.

Edit2: When using UTC DateTime by invoking ToUniversalTime() the performance is restored and clocks in at 1.669ms.