简体繁体 English

与直接数据库调用/RESTful 服务调用相比，对 Hazelcast 的数据检索速度进行基准测试的最佳方法是什么？

[英]What is the best way to benchmark Hazelcast's speed to data retrieval, comparing with direct DB calls / RESTful service calls?

原文 2020-08-22 21:23:15 0 1 java/ spring-boot/ hazelcast

Background: My team has been maintaining a mix of classic ASP/ASP.NET apps for over a decade.背景：十多年来，我的团队一直在维护经典 ASP/ASP.NET 应用程序的组合。 My company recently had a re-org in which teams and managers were either laid off or swapped around.我的公司最近进行了重组，其中团队和经理要么被解雇，要么被调换。 My new manager is pushing for a refactoring of our entire app portfolio over the coming year or two to Java/Spring-boot to meet company infrastructural standards.我的新经理正在推动在未来一两年内将我们的整个应用程序组合重构为 Java/Spring-boot，以满足公司的基础设施标准。 Most of our apps today perform ETL tasks via sprocs from MSSQL or Oracle SQL databases, with a few connecting to APIs via SOAP and REST respectively.我们今天的大多数应用程序通过来自 MSSQL 或 Oracle SQL 数据库的 sprocs 执行 ETL 任务，少数应用程序分别通过 SOAP 和 REST 连接到 API。

Goal: As part of this rewrite, I've been asked to help make a decision about how we handle data retrieval for our apps and to determine if there are any opportunities for improvement.目标：作为这次重写的一部分，我被要求帮助决定我们如何处理我们的应用程序的数据检索，并确定是否有任何改进的机会。 I want to try out Hazelcast (server) and "prove" it out-performs traditional data-retrieval methods like direct DB / REST calls.我想尝试 Hazelcast（服务器）并“证明”它优于传统的数据检索方法，如直接 DB/REST 调用。

Question: What would be the best approach for achieving my goal (benchmarking Hazelcast)?问题：实现我的目标的最佳方法是什么（对 Hazelcast 进行基准测试）？ My initial idea is creating a proof of concept with Java & Spring to test a.) the average speed of a direct DB call, b.) the average speed of a RESTful API call, and c.) the average speed of retrieval from Hazelcast.我最初的想法是用 Java 和 Spring 创建一个概念证明来测试 a.) 直接数据库调用的平均速度，b.) RESTful API 调用的平均速度，以及 c.) 从 Hazelcast 检索的平均速度. I would make sure the record set is consistent across all three tests.我会确保记录集在所有三个测试中都是一致的。 Is this the right approach, or am I over/under complicating it?这是正确的方法，还是我过度/不足使它复杂化？ Is there already something built out I can use to demonstrate the same?是否已经建立了一些我可以用来证明相同的东西？

(Note, I understand Hazelcast is not meant to replace the original source of record , but provide data caching for quicker retrieval of subsequent calls. I want proof that it is indeed quicker to retrieve from the cache rather than just doing another DB/API call). （注意，我理解 Hazelcast 并不是要替换原始记录源，而是提供数据缓存以便更快地检索后续调用。我想证明从缓存中检索确实更快，而不仅仅是执行另一个 DB/API 调用）。