dotnet/orleans

[PERF] Investigate optimizations for large graphs

Open

#8,140 建立於 2022年11月12日

在 GitHub 查看
 (1 留言) (1 反應) (0 負責人)C# (10,777 star) (2,123 fork)batch import
area-performancearea-serializationenhancementhelp wanted

描述

See this twitter thread for reference: https://twitter.com/neuecc/status/1589902671540256771?s=20&t=AcVRk8mwSdMyQJT2c5jyDg

Note: we could also consider avoiding reference tracking for types by default. There are some complexities involved, like what do you do with built-in types, eg string, byte[], List<object>. You could potentially have ref-tracking variants of the codecs or pass a reference tracking preference to the codec and have an attribute per-property to indicate a reference tracking preference.

貢獻者指南

技術棧
csharp
領域
performancebackend
議題類型
performance
難度面向新貢獻者的預計實作難度，1 表示很小改動，5 表示專家級工作。
4
預計時間有經驗貢獻者完成調查、實作、測試並準備 pull request 的粗略時間範圍。
over 1 week
活動狀態議題目前的可參與程度：新鮮、活躍、陳舊、阻塞或等待維護者輸入。
stale
清晰度議題是否清楚說明預期改動、驗收標準和下一步。
needs investigation
前置要求
C# knowledgeOrleans framework basicsserialization conceptsperformance profiling
新手友善度1-100 的估計分數，表示該議題對首次貢獻者的友善程度。
25
研究方向
Investigate the referenced Twitter thread by neuecc for optimizations used in C# serialization, particularly for large graphs. Consider avoiding reference tracking by default for types like string, byte[], List<object>. Explore creating ref tracking variants of codecs or passing a reference tracking preference to the codec with an attribute per property. Examine existing Orleans serialization code, likely in the Orleans.Serialization namespace, to identify current implementation details and potential optimization points.