专栏首页Jerry的SAP技术分享SAP CRM的订单模型移植到S/4HANA之后,到底做了哪些改进?

SAP CRM的订单模型移植到S/4HANA之后,到底做了哪些改进?

Overall idea One order model consists of a series of objects with two different types:

(1) extension: data segments which exclusively belong to the header or to an item. We therefore distinguish header extensions and item extensions. (2) set: a collection of data which can be linked to the header OR to an item of the transaction.

In the old world, each extension / set has each own persistence table:

And in the runtime each extension / set has their own runtime buffer as well, to store latest information modified by end user.

In new world, all fields from original extension / set are now merged into a single header and item table. Now for each transaction category like BUS2000116(Service Order), only one header and one item table are there.

(1) read: the set / extension data is read from new DB, and filled to corresponding object buffer in the runtime (2) write: the latest change splited in set / extension is merged into a single record, and persisted to new DB. Such work is done by convertor class for each set / extension.

We use github repository to manage issues & discussion during our development. 89 internal bugs are detected and fixed:

Phase 2 - CDS view development

The current Order advanced search in WebUI is implemented by advanced search:

CDS view design:

Check this excel: webui_search_fields_service

Phase 3 - prepare test data for performance test

Latest status on 2017-06-03 15:09PM

My progress

On Monday, I adapt Carsten’s test report. For the original implementation, if we specify a very large number of Service Orders to be created, say 1 million:

I will meet with this out of memory exception:

So I change the order creation behavior, from creation all orders in a single call to the creation using WHILE LOOP:

After that I perform a performance comparison, no performance loss after adaptation ?

And then I spend quite a lot of effort for creation optimization:

We are still far away from the target 100 million test data, so we have to use the clone way:

Phase 4 - Authorization integration to CDS view

check this excel: DCL_Mapping.xlsx A draft performance measurement is done, the result is no performance punishment after DCL is added to a CDS view. Feasibility study to use CDS view to re-implement CRM Interactive Report

Phase 5 - performance measurement using Gerwens, Heiko’s test framework - in process

本文参与腾讯云自媒体分享计划,欢迎正在阅读的你也加入,一起分享。

我来说两句

0 条评论
登录 后参与评论

相关文章

  • 使用Chrome开发者工具研究JavaScript里函数的原生实现

    As the size of my blog Chrome Development Tool tips used in my daily work turns ...

    Jerry Wang
  • How to create extension project from standard CRM Fiori application

    Pre-Steps: You should know how to run CRM Fiori 1.0 applications in WebIDE first...

    Jerry Wang
  • 关于面向对象设计中类的方法是否应该使用boolean类型的参数

    版权声明:本文为博主汪子熙原创文章,未经博主允许不得转载。 https://jerry.bl...

    Jerry Wang
  • 利用迭代细化进行依存关系语法分析的递归非自回归图到图转换器(CS and Language

    我们提出了一种通过非自回归图到图转换器的递归应用程序对任意图进行迭代细化的递归非自回归图到图转换器(RNG-Tr)。虽然之前自回归图预测中已经使用了\newci...

    用户6868260
  • 【Codeforces】1213A - Chips Moving

    版权声明:本文为博主原创文章,遵循 CC 4.0 BY-SA 版权协议,转载请附上原文出处链接和本声明。 ...

    喜欢ctrl的cxk
  • 资源 | Facebook开源DrQA的PyTorch实现:基于维基百科的问答系统

    选自GitHub 机器之心编译 参与:Panda 今年 4 月,斯坦福大学和 Facebook 人工智能研究所在 arXiv 发布了一个基于维基百科的开放域问题...

    机器之心
  • 研发:What is a DDoS Attack?

    A distributed denial-of-service (DDoS) attack is a malicious attempt to disrupt ...

    heidsoft
  • 神经网络零空间分析的异常值检测 (CS)

    许多机器学习分类系统缺乏能力意识。具体而言,许多系统缺乏识别何时将异常值(例如,不同于训练数据分布且未在训练数据分布中表示的样本)呈现给系统的能力。检测异常值的...

    DDDDDaemon
  • 一种改进新产品销售预测的网络转移学习方法(CS LG)

    数据驱动方法——例如机器学习和时间序列预测——被广泛用于食品零售领域的销售预测。 然而,对于新推出的产品,缺乏足够的训练数据来训练准确的模型。 在这种情况下,实...

    用户7095611
  • 原创译文 | 难民身份认证难?区块链来解决

    本文为灯塔大数据原创内容,欢迎个人转载至朋友圈,其他机构转载请在文章开头标注:“转自:灯塔大数据;微信:DTbigdata”

    灯塔大数据

扫码关注云+社区

领取腾讯云代金券