存储 + 评估绩效数据
由于我们的网络应用程序逐渐退化，我们决定监控我们的应用程序性能并衡量个人操作。
例如，我们将测量每个请求的持续时间、编辑客户或创建约会、搜索合同等单个操作的持续时间。 在大多数情况下，数据库是这些操作的瓶颈。
我预计最终数据将相当大，因为我们将针对每个请求收集 1-5 个单独的操作。
当然，将每个元素插入数据库是无意义的，因为这会进一步减慢每个请求的速度。
存储和评估这些每个请求数据的好策略是什么？ 我考虑过有一个附加的全局队列对象和一个清空队列并处理持久存储/文件的单独线程。 但这些数据存储在哪里呢？ 是否有任何用于此类可视化的预构建工具？
我们使用java、spring、混合hibernate+jdbc+pl/sql、oracle。
不过，这个问题应该与语言无关。
编辑：测量将在生产中进行很长一段时间。
since we suffer from creeping degradation in our web application we decided to monitor our application performance and measure individual actions.
for example we will measure the duration of each request, the duration of individual actions like editing a customer or creating an appointment, searching for a contract.
in most cases the database is the bottleneck for these actions.
i expect that the culminated data will be quite large, since we will gather 1-5 individual actions per request.
of course it would be nonsense to insert each an every element to the database, since this would slow down every request even more.
what is a good strategy for storing and evaluating those per-request data.
i thought about having a global Queue object which is appended and a seperate thread that empties the queue and handles the persistent storage/file. but where to store such data? are there any prebuilt tools for such a visualisation?
we use java, spring, mixed hibernate+jdbc+pl/sql, oracle.
the question should be language-agnostic, though.
edit: the measurement will be taken in production over a large period of time.
如果你对这篇内容有疑问，欢迎到本站社区发帖提问 参与讨论，获取更多帮助，或者扫码二维码加入 Web 技术交流群。
绑定邮箱获取回复消息
由于您还没有绑定你的真实邮箱，如果其他用户或者作者回复了您的评论，将不能在第一时间通知您！
发布评论
评论（1）
看来您的归档策略至少部分取决于您的测试范围：
至于可视化工具，我发现 Excel 对于少量到中等数量的数据非常有用。
It seems like your archive strategy will be at least partially dependent on the scope of your tests:
As for visualization tools, I've found excel to be pretty useful for small to moderate amounts of data.