This is a wonderful world

去自己想去的地方,呼吸那里的空气,贴近那里的土地,认识那里的人们......

导航

统计

公告

置顶随笔 #

[置顶]Japan Dragon Project Phase1 Lesson and Learned

摘要: We madea verygood improvement and did agood project compare to previous, But we still on the way to be a wonderful dev team. Following is project's lesson and learned.1. Planning Phase:a) OurBA did a ...阅读全文

posted @ 2007-07-19 11:59 shyuan 阅读(412) 评论(0) 编辑

[置顶]系统性能分析(找出trouble maker)

摘要: 系统性能分析(找出trouble maker): 老听到客户抱怨“你的系统很慢啦!”做为无所不能的设计者,该这么解决这个问题呢? 阅读全文

posted @ 2007-06-11 17:02 shyuan 阅读(3403) 评论(12) 编辑

[置顶]团队建设(4)-天下没有不散的筵席

摘要: 铁打的营盘流水的兵,团队成员们相聚一场,最终难免要离开。有的成员羽翼日渐丰满,自然要高升/跳槽;达不到考核底线的成员也需要另谋出路。做一个开发组长,最得意的莫过于当其他部门成立开发组时,他们的老板跑过来问你“你们有没有人可以到我们那领导一个开发组的?”当这种好事来时,你找合适的组员私下谈谈,谈谈这个职位的要求和他的职业生涯规划(这也是为什么团队建设(3)中一直大力鼓吹生涯规...阅读全文

posted @ 2007-03-31 12:13 shyuan 阅读(2307) 评论(21) 编辑

[置顶]团队建设(3)-组员职业生涯规划

摘要: 看完DVD“肖恩克的救赎”,感受颇深:一个人不能没有希望,更不能没有目标。想想组员们为什么能接受我安排的工作,受我影响,认为我们是他们的Mentor(导师)?因为我们能帮助他们成长,在他们通往职业目标的途中帮助他们。一个人愿为公司奉献青春,无非为了两点:要不能赚到钱,要不能学到本事。而开发组长呢,能控制的财力要不没有,要不很少。大多数的开发人员都很聪明,愿意为了明天而投资。...阅读全文

posted @ 2007-03-27 00:32 shyuan 阅读(2983) 评论(22) 编辑

[置顶]团队建设(2)-对评论的回复和具体操作1

摘要: 如果“负责任”要成为团队的价值观,那开发组长呢?当以身作则,成功实施项目和培养优秀组员是一个称职开发组长的使命。“没有办法,这是中国国情,一切精神的发展都要在物质基础上,等我有了500万,我会好好培养一支团队,但是我没有。”这样的理由是否能成为放弃使命的借口呢?作为团队的带头人,请对各种各样的理由说“NO”!诚然,刚开始会碰到...阅读全文

posted @ 2007-03-24 01:36 shyuan 阅读(2648) 评论(9) 编辑

[置顶]团队建设(1)

摘要: 又一个项目成功上线了。老板说:干的不错!你满心欢喜的期待提拔呀,钞票大大的涨呀......但迟迟没有动静?忍不住旁敲侧击,被老板一语道破:保证项目成功是开发组长的职责,你的能力不错,但现在无法提拔你,因为团队成员中没人可以接班。有的项目成功,很大程度上取决于开发组长的个人能力。如果开发组长技术能力和项目管理能力超强,带领着一群普普通通的开发人员,一个7×24的生产系统一样炮制出来。但这就...阅读全文

posted @ 2007-03-22 22:22 shyuan 阅读(17655) 评论(37) 编辑

2007年7月19日 #

Japan Dragon Project Phase1 Lesson and Learned

 

We made a very good improvement and did a good project compare to previous, But we still on the way to be a wonderful dev team. Following is project's lesson and learned. 

1. Planning Phase:
a) Our BA did a good job in business analysis, So some of developers start work without fully understand the whole business process, only focuses on his function. Though it isn't impact project successful. But it holds back team members improve business concept and to be talent developers. (A good technical skill & a fast business understanding comprise a talent developer) The reason of  business concept shortage is "clear individual working/function scope and lack of interesting" after questioned each team member. "Clear function scope" is right for a big project. Shall focuses on how to anti-"lack of interesting".
b) New hire don't fully understand his job function, made more mistakes compare to other team members. "Correct action base on correct understanding". How to improve business function understanding?

Solution:
1. Emphasize a concept that is "master whole system business process is a key talent as a team leader and an architector". Make decision by yourself.
2. Add brief function introduce in peer review process.
3. Re-draw business processing diagram again after communicated with business analyst. Email his diagram to BA to double confirm. Email content shall contain all function related information, (Exp: If one function point changed, email content shall include whole function processing diagram and high light the changes).
4. New hire's fist mail shall be helped and sent by his mentor. Mentor give him/her a good example first.
5. Add more check points for new hire's job function.

2. Development Phase
a) Peer review wasn't enough, only covered 70% code.
b) System configuration met a little trouble. Whole team was continue struggling on the "right" configuration during the SIT phase until we made an agreement using same entry point to change system configuration.
c) We successfully did every day's "Daily Build", but don't have time to do "Daily deployment". 
d) Sometimes dlls updated wrongly in SIT phase. There are successfully build dlls in daily build output folder, why not get them? That means urgent working time system update shall trigger daily build, passed build first.
e) Load test only covered 70% user case. Not too bad, not too good.

Solution:
1. A team member volunteer to be a peer review
moderator. Trying to find a best practice balancing time and efficiency. She will help to plan team members time, follow up working item and so on.
2. Use same entry point for system configuration. Four entry points are required for Dev, SIT, UAT and Production configuration.
3. Import "Daily Deployment" in next project.
4. Get suitable dlls from daily build folder for every SIT, UAT deployment.
5. Load testing is a time consuming work item, key scenarios cover is required, all scenarios cover is better.

3. Stabilization Phase
a) Backlog orders problem. We did a little preparation for backlog orders which shall be loaded into new system. In the system go live day, backlog orders gave us a big surprise, there were too many of them, out of system capacity. Have to temporarily wrote a little program to split orders, it took 3 hours. 
b)11 issues counted in the first two weeks, 3 of them belong to careless issue. Careless is a problem for some young developers, They will grow up by tasting and tasting painful result for his/her careless. It need time, What team can do is improve peer review process.

Solution:
1. Smoothly pass backlog data trap is not too difficult: a)How many of them? b) How to process it? c)Test it.

Enhanced peer review steps:
1.Follow up last review agreement.
2.Code owner hold brief function introduction, and present actual function.
3.Code owner run unit test
4.Reviewer review code sequence diagram
5.Reviewer go through unit test code and code.

Task completed standard:
In what kind of situation we can say "Task completed! Job done!" in Monday meeting during development phase?
1.Function completed
2.Unit test code coverage up to "90%"
3.Peer review passed 

posted @ 2007-07-19 11:59 shyuan 阅读(412) 评论(0) 编辑

  下一页