[歌词] 希望和悲伤

看不清，你那闪烁的眼睛
听不清，你那隐约的耳语
你的心，仿佛忽远又忽近
我的心，猜不透你我的结局

慢慢靠近，缩短你我的距离
依然小心，避免莫名的抗拒
回首往事，你曾开心地鼓励
搜寻记忆，欢笑陪伴着泪水

而泪水，记录了一天又一天
而欢笑，憧憬着一夜又一夜

希望和悲伤只是我的幻觉
在梦里依然期盼你的出现
不要说过去只是美丽的错误
就让一切留待明天去回忆

希望和悲伤不是我的幻觉
在梦里依然看见你的出现
如果说过去只是美丽的错误
就让一切留待明天去回忆

The 5S Rules in IT System Design

Here 5S Rules means: Simple, Solid, Strong, Safe, Smart.

Simple
Simple means: 1. the logic is easy to understand; 2. the architecture has weak coupling connections.

According to Modular Design methodology, generally during design, we firstly define the Domain of the system, and then define Modules within system, and then define Function Points within Module.

When wo do Top-Down design, it's easy to follow Simple Rule, but sometimes the Function grows slowly, it extends from a Function Point to a whole system, in that case, we need to reconstruct the system design, and that's painful.

Let's take an example of Manufacturing System.

There's a work station in the workshop, which needs a worker to do assemlby and quality check after that. So strictly speaking, this worker combins 2 roles: Operator and Quality-checker. But since it requires low skill level and long cycle time, so these 2 functions have been combined to 1 single person. Accordingly, the IT system writes 1 single Logic for that.

But as it requires shorter cycle time and higher efficiency, new worker has been recruted, so this workstaton has been extended to 2 workstations shared with 2 persions, 1 for assembly, 1 for quality checking. At that point, IT system will also be required to split the funtion into 2 Function Points.

To reduce development time, we might add some IF THEN ELSE judgement into original logic, and then different role executes different logic block. That means, load the whole logic first, then execute its sub-logic-block.

Another solution is, analyze the request following Simple Rule even during original design, to package Assembly & Quality-check logic into different Logic Block, and then combine the Logic Block during deployment. That means, 1 person will be assigned with 2 roles' logic before change, and 2 persons will be assigned with their own roles with their own Logic Blocks after change. The background codes only covers Logic Block, and frontground will display it according to role-logicblock configuration.

Solid
Solid means: The logic need to be solid and complete.

The completeness of business logic could help to reduce system exception, and increse system availability.

An effective method of executing Solid Rule is to run Black-box test, let users to write test cases, let users to do the acceptance test.

Because developing guys focus on nornal work flow mostly, but users notice more exceptional work flow, and the test for exceptional work flow helps to increase the completeness of business logic.

Strong
Strong means the system needs to be strong, can support multiply sessions, lots of users, huge data. The system needs to reduce down time throughout all kinds of methods.

As for hardware, we may use Cluster, Load Balance and other technology to increse system availability.

As for software, we may use Cache, Read-write seperation, User grouping, Redundancy, Archive, Database Partition, and other technology to reduce computing time.

One simple method of keeping strong is to find the Bottle Neck of the problem. For example, the bottle neck of one system is running one specific report, which may cause long time waiting if several persons are running it in the same time. The resolution is to increase hardware performance to reduce query time, or to optimize query SQL and Table struchture, or to dispatch report with user permission, such as running report in advance during free time.

Safe
Safe means to avoid service interruptions by all means.

The Black Swan Inccidents - unpredictable exceptional inccidents - are difficult to avoid, the resolution is to accumulate problem resolving methods, and deal with exceptions by alert, logging, notification, so the system can continue to serve.

It will be good to desing a configurable Exception-handling Module, which could take exceptions quickly without rewriting core codes.

Smart
Smart means to use smart methods.

Now we have a lot of open source codes, modules, methodology, and a lot of internet appplications might be introduced into Enterprise systems.

For example, Map-Reduce is known as Google's methodology, it has been used in lots of web sites, and actually we could use it to simplize logic blocks with a lot of IF THEN ELSE judgement.

IT系统设计中的5S原则

5S原则：Simple, Solid, Strong, Safe, Smart.

Simple
Simple就是简单原则，这里的简单有两个意思，一个是逻辑简单容易理解，另一个是架构简单耦合性弱。
按照模块化设计的方法，通常我们在设计时，首先界定系统的边界，然后在系统内划分模块，然后在模块内划分功能点。
通常在进行自顶向下设计的时候，Simple原则是很容易遵守的，但是有时功能是在慢慢增强的，从一个功能点成长为一个系统，在后期开发的时候往往会对前期的设计进行重构，这是相当痛苦的。
举一个制造系统的例子。
比如车间有一个工位，要求工人做装配的活，装配完成后做质量检查。严格地说，这个工人集成了两个角色：装配工和质检员，但是由于生产技术难度低，对生产节拍要求也低，因此为了节省人力成本，这两个动作由同一个人完成。而对应的IT系统功能将逻辑也统一起来，以方便用户操作。
但是后来生产节拍要求越来越高，为了提升效率，车间招募了新的工人，一个工位变成了两个工位，一个人的活变成了两个人的活，其中一人专门做装配，另一人专门做质检。这时要求IT系统将功能拆分出来，将一个逻辑变成了两个逻辑。
通常为了减少开发的时间，我们可以在原先的完整逻辑中增加若干个IF THEN ELSE的判断，然后让不同的角色执行不同的逻辑块。也就是说，先加载逻辑，再判断执行。
另一种设计思路是，在第一次开发的初期，按照Simple原则进行分析，将装配的逻辑和质检的逻辑分别封闭成逻辑块，在实施的时候根据配置的情况进行组合。对应的，变更前一个用户执行两个角色的逻辑，变更后两个用户各执行对应角色的逻辑。后台的代码只负责各自独立的逻辑，前台的展现则根据配置的情况对逻辑进行组合。

Solid
Solid就是逻辑要牢固完整。
业务逻辑的完整可以减少系统的异常，从而提升系统的可用性。
执行Solid原则的一个有效方法是黑箱测试，让用户撰写测试场景，让用户进行接收测试。
开发人员往往想到正常的业务流，而用户则熟悉更多异常的业务流，而对异常业务流的测试可以提升逻辑的完整性。

Strong
Strong就是系统要可靠，能够支持多并发性、多用户、大数据量，通过各种方式减少当机时间。
硬件方面，可以通过Cluster、Load Balance等提升系统的可用性。
软件方面，可以通过缓存、读写分离、用户分组、冗余、归档、数据库分区等技术减少计算时间。
保持Strong的一个简单原则是找到并解决瓶颈所在。比如某个系统的瓶颈是跑某个特定的报表，如果几个人在同一个时间段跑此报表，则造成数据库查询长时间等待。解决的办法是从硬件的角度提升数据库的查询效率，或者优先查询语句及表结构，还有一种办法是在用户允许的情况下，对查询请求进行调度，比如提前在空闲的时间段事先在后台执行查询。

Safe
Safe就是安全原则，通过各种办法避免系统服务的中断。
在IT系统中，黑天鹅事件(无法预测的异常事件)是难以避免的，解决的办法是不断累积异常处理措施，每当有异常的时候，通过报错、日志、通知等方式将异常处理，从而避免系统服务的中断。
系统在设计的时候，应该规划一个可配置的异常处理模块，这样可避免在重写核心代码的情况下实现异常的及时处理。

Smart
Smart就是聪明原则。
现在网络上开源的代码、模块、方法论很多，许多互联网的应用也可以借用到企业系统。
比如Map-Reduce是GOOGLE倡导的方法论，实际上许多包括大量IF THEN ELSE的代码都可以利用Map-Reduce进行逻辑的简化。