Skip to content

.NET 7: Kestrel performance improvements on high core machines #26745

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
tdykstra merged 4 commits into from
Aug 16, 2022
Merged

.NET 7: Kestrel performance improvements on high core machines #26745

Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
2204ab2
kestrel perf on high core machines
tdykstra Aug 15, 2022
7bd5544
fix capitalization
tdykstra Aug 16, 2022
acb48f6
Merge branch 'main' into net7kestrelperf
tdykstra Aug 16, 2022
72fc215
Update aspnetcore/release-notes/aspnetcore-7.0.md
tdykstra Aug 16, 2022
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
8 changes: 8 additions & 0 deletions aspnetcore/release-notes/aspnetcore-7.0.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -187,6 +187,14 @@ One place where these improvements can be noticed is in gRPC, a popular RPC fram

![Entity diagram](https://user-images.githubusercontent.com/219224/177910504-e93579b4-02e4-4079-8a8c-d9d24857aabf.png)

### Kestrel performance improvements on high core machines

Kestrel uses <xref:System.Collections.Concurrent.ConcurrentQueue%601> for many purposes. One purpose is scheduling I/O operations in Kestrel's default Socket transport. Partitioning the `ConcurrentQueue` based on the associated socket reduces contention and increases throughput on machines with many CPU cores.

Profiling on high core machines on .NET 6 showed significant contention in one of Kestrel's other `ConcurrentQueue` instances, the `PinnedMemoryPool` that Kestrel uses to cache byte buffers.

In .NET 7, Kestrel's memory pool is partitioned the same way as its I/O queue, which leads to much lower contention and higher throughput on high core machines. On the 80 core ARM64 VMs, we're seeing over 500% improvement in responses per second (RPS) in the TechEmpower plaintext benchmark. On 48 Core AMD VMs, the improvement is nearly 100% in our HTTPS JSON benchmark.

## Server

### New ServerReady event for measuring startup time
Expand Down