Rapid Avro serializer for C# .NET
Avro format combines readability of JSON and compression of binary data serialization.
The main purpose of the project was to enhance HTTP communication between microservices. Replacing JSON with Avro brought three main benefits:
- Decreased the communication time between microservices
- Reduced the network traffic by about 30%
- Increased communication security - the data was not visible in plain JSON text
| AvroConvert | Apache.Avro | Newtonsoft.Json | |
|---|---|---|---|
| Rapid serialization | ✔️ | ✔️ | ✔️ |
| Easy to use | ✔️ | ❌ | ✔️ |
| Built-in compression | ✔️ | ✔️ | ❌ |
| Low memory allocation | ✔️ | ✔️ | ✔️ |
| Support for C# data structures: Dictionary, List, DateTime... | ✔️ | ❌ | ✔️ |
| Support for compression codecs | Deflate Snappy GZip Brotli |
Deflate | ❌ |
| Readable schema of data structure | ✔️ | ✔️ | ✔️ |
| Data is encrypted | ✔️ | ✔️ | ❌ |
Results of BenchmarkDotNet:
| Converter | Request Time [ms] | Allocated Memory [MB] | Compressed Size [kB] |
|---|---|---|---|
| Json | 672.3 | 52.23 | 6044 |
| Avro | 384.7 | 76.58 | 2623 |
| Json_Gzip | 264.1 | 88.32 | 514 |
| Avro_Gzip | 181.2 | 75.05 | 104 |
| Json_Brotli | 222.5 | 86.15 | 31 |
| Avro_Brotli | 193.5 | 74.75 | 31 |
Article describing Avro format specification and benchmark methodology: https://www.c-sharpcorner.com/blogs/avro-rest-api-as-the-evolution-of-json-based-communication-between-mic
Conclusion:
Using Avro for communication between your services significantly reduces communication time and network traffic. Additionally choosing encoding (compression algorithm) can improve the results even further.
- Serialization
byte[] avroObject = AvroConvert.Serialize(object yourObject);- Deserialization
CustomClass deserializedObject = AvroConvert.Deserialize<CustomClass>(byte[] avroObject);- Read schema from Avro object
string schemaInJsonFormat = AvroConvert.GetSchema(byte[] avroObject)- Deserialization of large collection of Avro objects one by one
using (var reader = AvroConvert.OpenDeserializer<CustomClass>(new MemoryStream(avroObject)))
{
while (reader.HasNext())
{
var item = reader.ReadNext();
// process item
}
}- Generation of C# models from Avro file or schema
string resultModel = AvroConvert.GenerateModel(_avroBytes);If you find my work useful, you can support it :)
The project is CC BY-NC-SA 3.0 licensed.
For commercial purposes purchase AvroConvert on website - Xabe.net
We want to improve AvroConvert as much as possible. If you have any idea, found next possible feature, optimization opportunity or better way for integration, leave a comment or pull request.