Skip to content
View DZQOX's full-sized avatar

Block or report DZQOX

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
DZQOX/README.md

Hi, I'm zq 👋

Engineer at Softprobe · softprobe.ai · zq@softprobe.ai

Working on the Java agent — bytecode instrumentation, production traffic record & replay, AI-driven diff diagnosis.


Softprobe — Know Your Software

AI-powered production-traffic record-and-replay testing, for enterprise-scale Java systems.

English · 中文 · softprobe.ai · info@softprobe.ai


English

We are the team behind AREX — the open-source traffic record-and-replay project born inside Trip.com's million-QPS core systems. Softprobe is the enterprise edition of that work: broader coverage, tighter overhead, AI diagnosis, and a professional team on the ground.

What we do

Capture real production traffic at the JVM bytecode level, replay it in an isolated environment with all downstream dependencies auto-mocked, and let AI diff the results and point straight at the root cause.

  • Zero code intrusion — one -javaagent flag, no source changes
  • Zero environment setup — HTTP / SQL / Redis / MQ / RPC all auto-mocked from recording
  • Real-path coverage — production traffic is the test corpus; no more hand-written cases
  • AI-driven diagnosis — noise fields (timestamps, nonces, encrypted blobs) auto-filtered; true diffs get a root-cause pointer

Where it lands

  • Regression testing — record once, auto-replay on every release
  • Canary / grey release validation — validate the new build against real traffic in isolation, without risking real users
  • Database migration — Oracle / MySQL → OceanBase / DaMeng / GaussDB, replayed end-to-end so business correctness (not just data) is verified

Battle-tested

  • Million-QPS core systems since the Trip.com days
  • 4,100+ applications onboarded
  • 150M+ recorded requests per day
  • Production customers include a top-tier insurance carrier (Oracle → OceanBase core replacement) and a global travel platform

What the commercial edition adds on top of AREX

Open-source AREX Softprobe
Instrumentation modules ~30 68+ (JDBC · iBatis · Spring Boot 2/3 · Jakarta · Nacos · SOFA RPC · gRPC · domestic gateways)
Overhead control Sampling & throttling CPU +<5% verified at million-QPS; auto-degrade on threshold
Diagnosis Manual diff review AI noise filtering + root-cause localization
Coverage adaptation Community Dedicated on-site engineering team

Learn more


中文

我们是 AREX 开源项目的原班团队 —— 先在携程百万级 QPS 核心系统长期验证,再沉淀为流量录制回放的开源项目 AREX,如今打造为企业级商业产品 Softprobe

我们做什么

JVM 字节码级 采集真实生产流量,在隔离环境回放,下游依赖自动 Mock,AI 自动对比差异 + 定位根因

  • 零代码接入 —— 加一行 -javaagent,业务代码零改动
  • 环境免准备 —— HTTP / SQL / Redis / MQ / RPC 全自动 Mock,无需搭环境、造数据、维护 stub
  • 真实路径覆盖 —— 用例由真实流量自动沉淀,覆盖人工想不到的调用路径
  • AI 自动诊断 —— 时间戳、随机数、加密串等噪音字段自动过滤,真差异 AI 直接指到根因

落到哪里

  • 回归测试 —— 录一次真实流量,每次发版自动回放比对
  • 灰度发布验证 —— 隔离环境用真实流量验证新版本,不必拿真实用户冒险
  • 数据库迁移 —— Oracle / MySQL 迁到 OceanBase / 达梦 / GaussDB,不止比数据,更比业务逻辑

千锤百炼

  • 百万级 QPS 核心系统验证规模
  • 4100+ 已接入应用
  • 1.5 亿次/日 录制流量
  • 客户包括头部保险企业(Oracle → OceanBase 核心替换)、全球化出行平台等

商业版在开源之上,多了什么

开源 AREX Softprobe 商业版
插桩模块 约 30 68+(JDBC · iBatis · Spring Boot 2/3 · Jakarta · Nacos · SOFA RPC · gRPC · 国内网关等)
开销控制 采样限流 百万 QPS 生产环境实测 CPU 增加不足 5%;超阈自动降级停录
诊断能力 人工比对 AI 噪音过滤 + 根因定位
覆盖适配 社区维护 专职团队现场适配

了解更多


Know Your Software · 用真实生产流量,让每一次上线都有底气

Popular repositories Loading

  1. EICA-based-on-Zigbee EICA-based-on-Zigbee Public

    基于zigbee的智能监控系统

  2. test1 test1 Public

    2016/08/03 23:18

  3. Hello-Word Hello-Word Public

  4. test test Public

  5. learngit learngit Public

  6. learngit2 learngit2 Public