Skip to main content

Steam.CopyTo - Performance problems

This week I heard about an interesting performance problem, that I want to share with you.
There is a desktop application that is used to import data from files to database. These files were in different formats, from XML to binary. Everything was fine, the application work great in the development and first phase of testing. All the code was double check, each stream was dispose when it was no longer used.
During the testing phase, the application started to have an odd behavior, the RAM memory started to go crazy. Normally, the application use around 200MB during an import, but there were times when more than 1500MB of ram were consumed. Not only this, but “Out of memory” exception appeared a lot of time during different imports. All this started too appeared at the testing phase.
With the same setup, developers manage to reproduce the problem and using the memory profiles that comes with Visual Studio the problem was spotted.
Stream stream1 = …
Stream stream2 = …
stream2.CopyTo(stream1);
The problem was with CopyTo method. Because this method was called a lot of time in a short period of time, the GC didn’t had time to dispose all the resources used during the copy process. This can become a big problem when the size of stream is pretty big. This method was introduce in .NET 4.0 and the scope of it is to copy the content of one stream to another. This is done by reading to a buffer of bytes (default buffer size is 4096) chunks from the stream and write them to the second stream. 
It is very similar with what we usually do in .NET 3.5 when we had to copy one stream to another:
public void CopyStream(Stream input, Stream output)
{
    byte[] buffer = new byte[size];
    int read;
    while ((read = input.Read(buffer, 0, buffer.Length)) > 0)
    {
        output.Write (buffer, 0, read);
    }
}
The problem is with the buffer that is represented as an array of bytes. If we call the CopyTo method from time to time with small streams will not have any kind of problem. But when we use big streams intensive, calling this method very often can create huge problems. Each time when we make new calls to this method, a new part of our memory will be allocated for the buffer. GC will not have time to clean the memory and we can end up with a lot of memory that is not disposed.
The solution for this problem was pretty simple. The CopyTo method was replaced with a BinaryReader that was used to copy the content of stream to an array of bytes. This array of bytes was later used to create the MemoryStream. In this way, the stream that was used for read data is disposed immediately (BinaryReader) and the array of bytes that is used as a buffer is used directly in the memory stream.
byte[] bytes;
using( BinaryReader reader = new BinaryReader(inputStream))
{
    bytes = reader.ReadBytes((int)inputStream.Length); 
}
outputStream = new MemoryStream(bytes, 0, bytes.Count());
Using this solution, the problem with the memory leak disappeared. It seems that not all time, the new API is the best solution when we need performance.

Comments

  1. What if size of input stream will be too big?
    it will try to create huge bytearray, and possible OutOfMemoryException - no?

    ReplyDelete
  2. Wow, thanks, this helps alot !

    ReplyDelete

Post a Comment

Popular posts from this blog

Why Database Modernization Matters for AI

  When companies transition to the cloud, they typically begin with applications and virtual machines, which is often the easier part of the process. The actual complexity arises later when databases are moved. To save time and effort, cloud adoption is more of a cloud migration in an IaaS manner, fulfilling current, but not future needs. Even organisations that are already in the cloud find that their databases, although “migrated,” are not genuinely modernised. This disparity becomes particularly evident when they begin to explore AI technologies. Understanding Modernisation Beyond Migration Database modernisation is distinct from merely relocating an outdated database to Azure. It's about making your data layer ready for future needs, like automation, real-time analytics, and AI capabilities. AI needs high throughput, which can be achieved using native DB cloud capabilities. When your database runs in a traditional setup (even hosted in the cloud), in that case, you will enc...

How to audit an Azure Cosmos DB

In this post, we will talk about how we can audit an Azure Cosmos DB database. Before jumping into the problem let us define the business requirement: As an Administrator I want to be able to audit all changes that were done to specific collection inside my Azure Cosmos DB. The requirement is simple, but can be a little tricky to implement fully. First of all when you are using Azure Cosmos DB or any other storage solution there are 99% odds that you’ll have more than one system that writes data to it. This means that you have or not have control on the systems that are doing any create/update/delete operations. Solution 1: Diagnostic Logs Cosmos DB allows us activate diagnostics logs and stream the output a storage account for achieving to other systems like Event Hub or Log Analytics. This would allow us to have information related to who, when, what, response code and how the access operation to our Cosmos DB was done. Beside this there is a field that specifies what was th...

[Post Event] Azure AI Connect, March 2025

On March 13th, I had the opportunity to speak at Azure AI Connect about modern AI architectures.  My session focused on the importance of modernizing cloud systems to efficiently handle the increasing payload generated by AI.