追求卓越，勇攀高峰 - RWP中国之旅盛大来袭

Andrew Holdsworth

Oracle Real-World Performance 全球副总裁

It’s 20 Mar, only a week before the RWP China Tour, another technical seminar about Oracle database will also be held at the end of this month in Beijing, on cooperation with RWP and ACOUG.

由RWP团队和ACOUG联合主办的RWP 中国之旅将于本月28在京隆重举办，只剩下一个星期，一场围绕Oracle性能的技术盛会将全面袭来，是不是很期待？

We are all so excited about this conference, and the other reason we are gathering here today is (now you can prepare your applause if you like) that I am going to introduce you Mr. Holdsworth, the Vice President of Real World Performance at Oracle!

今天，还有一个更加让人振奋的消息要带给大家！我们有幸邀请到了RWP团队全球副总裁Andrew Holdsworth，在会前跟大家见面！以下是对Andrew的采访。

Q1：Hello, Mr. Holdsworth. It’s a honor to have this conversation with you, I guess most people work on database have known you well, but for new friends who might also read this, please tell something about yourself, say hello to everybody.

问题一：Holdswort先生您好，我相信在座的Oracle数据库技术的爱好者都对您很熟悉，当然可能还有一些行业的新朋友，请您简单自我介绍以下，跟大家打个招呼吧？

Andrew ：Hello, I have been working for Oracle now 27 Years, Not as long as Graham but probably longer than some of our Chinese customers have been alive ! I have been working on Oracle performance projects since my second week at Oracle.

大家好，我在Oracle行业已经工作了27年了，虽然没有Graham的时间长，但可能比我们在中国的某些客户存活的时间还要久，事实上在我进入Oracle的第二个星期开始，我就全身投入到Oracle性能优化的工程中了。

I never intended to make performance my career but I just kept being asked to help with performance problems and I guess my reputation started to build. However over the 27 years I have been very lucky to work and learn from very talented engineers all over the world so I cannot claim all the credit for my reputation. I would however advise all young engineers to find good mentors and role models to help them through their careers. The subject is so big you cannot know everything. Sometimes it is just better to know who to ask!

最初我并没有打算在性能优化方向上做长远发展， 但经常有人找我帮忙处理此类的问题，久而久之，我逐渐因性能优化而出名。事实上在我27年性能优化的生涯中，我非常幸运能够与一群优秀的工程师共事，并互相学习一起进步，所以Oracle在性能优化上的提升和改进这不是我一个人的功劳，我经常会给团队里年轻的成员这样的建议，在工作中树立榜样，寻找目标，这样才能更快地进步。性能优化是一个很大的工程，你不可能搞懂所有的事情，有时候你只需要知道，该去请教谁，从而少走弯路。

Outside of work my passions are sailing and cycling. I have long stopped racing bicycles but I still race my boats as well having adventures sailing between countries. I try and Sail in Hong Kong Harbor when visiting China.

工作之外，我热衷于划船和骑车，现在已经很长时间没有跟伙伴们竞赛骑车了，但一直坚持划船，最喜欢划船从一个国家到另一个国家，这是一种有趣的冒险，这次去中国，我很想直接从香港划船过去。

Q2：It has been two years since last RWP China Tour, which was held in Oct 2015. No need to mention that it was a real great success, 300 technology enthusiasts gather together, they sharing, discussing and of course progressing. It was so much more than collision of technology. and now, with 28 Mar upon us, we will once again usher in such an important gathering. I can’t wait to share this exciting news with the Oracle database technology enthusiasts in China, so they can make their schedules to listen your speech in advance.

问题二：上一次的RWP中国之旅是在2015年举办的，那是一次成功的技术盛会。广大的Oracle技术爱好者欢聚一堂，思维碰撞，共同成长！3月28，我们将再次在北京相聚，我相信大家都已经迫不及待要再次一睹RWP团队的风采，并在其最新研究中感悟技术之美！

Mr. Holdsworth，Would you like to share some interesting memory or the most unforgettable part about last RWP China Tour in Beijing? Simultaneously, what’s your biggest expectation about this time, what subject will you share in conference?

Holdsworth先生，关于上次的RWP中国之旅您还记得吗？有没有什么有趣或者难忘的经历想和大家分享的呢？

这次RWP之行，您有什么期待，您将会分享什么内容给大家呢？

Andrew ：Yes it has been a while，I remember it very well. One day I was sailing in Hong Kong harbor inshorts and a Teeshirt the next day I was in a snow storm in Beijing!

上次RWP到中国的确已经很长时间了，但我仍然记忆尤新。我记得前一天我还穿着T恤和短裤在香港划船，第二天就在北京看暴风雪了！

It is always a thrill and a privilege to visit China. I love the size of the database projects, the energy and youth of the technical teams my Team and I work with in China.

来中国对我来说是一种荣幸，也是一件刺激的事情。我们在中国的团队和项目很大，中国区的同事充满了年轻的能量，我们也努力将中国区的项目做得更大更好。

This year I hope to see some familiar faces, friends , meet new friends all who have a passion for making their systems go faster and wanting to fully exploit the power of Oracle Database Technology.

本次去中国，我希望能够看到一些熟悉的老朋友，看到行业内的新朋友和任何对Oracle数据库性能的探索和优化充满热情的朋友们！

On this Trip Graham and I will demonstrate some the old traditional performance techniques in new innovative ways as well introduce some the new challenges we are seeingcoming out of Real World Database Applications.

在这次中国之旅，Graham和我将以创新的方式展示一些旧的传统性能技术，并介绍一些我们从真实世界数据库应用程序中看到的新挑战。

Q3：Mr. Holdsworth, I have learned about your Performance optimization philosophy,

（1）Exceptional performance has never been accidental

（2）Your system performance is what you "deserve"

（3）"Almost, good enough" has always been "a lot worse,not good enough", it is always necessary to pursuit of excellence.(please excuse me, for maybe it’s not exact words, pretty much the gist)

问题三：Mr. Holdsworth,我了解到您在做性能优化中所遵循的一些原则：

• 良好的性能从来都不是偶然；

• 你的系统性能是你期待它所成为的样子；

• “差不多，够好了”其实就是“差很多，不够好”。追求卓越永远都是必须的。

良好的性能从来都不是偶然；

你的系统性能是你期待它所成为的样子；

“差不多，够好了”其实就是“差很多，不够好”。追求卓越永远都是必须的。

Would you like to share more, how did you create and insist on this philosophy, and how does it make difference to your career?

关于您的优化原则，您愿意跟我们分享一下您当初是基于什么想法制定了这样的原则，这些原则在您的职业生涯中产生了哪些影响呢？

Andrew ：I guess I was taught at an early age it is better to do a few things really well rather than do a lot of things sub optimally. I apply this mindset to performance projects and life in generally. I really enjoy getting things right. I really love winning races in my boat!

我在很小的时候就被教导，样样精通其实会样样稀松，你必须找到你最擅长的事情，做到极致。我把这样的想法应用于我的工作和生活，在工作中，我坚持把性能优化这件事做到最好，在生活中，我全身心投入并享受划船和与之相关的竞赛。

Q4：The RWP team works on achieving the best performance of the systemin real world over these years. So based on your database performance tuningexperience,have you ever got stumped by any kind of problem, or a certain industry or system environment, does the performance tuning method differ fromeach industry?

问题四：在您多年做性能优化的过程中，有没有被哪一类型的问题，或者哪一行业，哪一特定系统环境下的问题难倒呢，在不同的行业，性能优化的方法有哪些区别呢？

Andrew ：All performance problems are a challenge and require a certain amount of puzzle solving skills. However across all geographies and industries we do see similar mistakes that hurt system performance on a consistent basis. These common mistakes are the basis of most Real World performancework and yield orders of magnitude increases in performance when corrected. It is not uncommon to get 1000 times the performance by applying real world performance techniques.

所有的性能问题都有不一样的特性，也需要因地制宜使用不同的技巧。不过在我们的工作经验中，的确看到很多相似的问题导致系统性能较差，而RWP大部分时候回基于这些错误点进行性能调整和优化，最终性能都能得到较大的提升。通过RWP的性能技术获得1000倍以上的性能提升并不罕见。

The biggest challenges we face are actually human/political. When you show people they have designed a suboptimal system or have used the database incorrectly there is a lot of defensive behavior and resistance to change andnew ideas. This is the hardest problem to solve.

我们面临的最大的问题事实上是人的问题或者管理策略的问题，当我们向系统管理者展示他们设计的次优系统或者使用数据库不正确时，他们很抵触，也不愿意改变。 这是最难解决的问题。

Q5：In the two main kinds of database environment, OLTP and Datawarehouse, when you do optimization, do the basic strategy and tuning methodsbe any different and how?

问题五：两种主要的数据库环境，数据仓库和OLTP，在进行性能优化的时候策略和方法有哪些不同？

Andrew ：performance usually focuses on execution of millions of very short SQL statements.In this case the SQL, the development environment and connection strategy has the most impact on performance. However in DW we are looking at low execution count of long running, resource intensive SQL often in parallel. For the DW focus isagain on SQL, data design and ensuring we can recruit all the available HW tosolve the DW query challenges.

性能通常专注于执行数以百万计的非常短的SQL语句。 在这种情况下，S开发环境和连接策略对SQL性能的影响最大。 然而，在DW中，我们正在寻找执行较少、运行时间长的SQL，资源密集型SQL往往并行执行。 对于DW重点关注的SQL，数据设计和确保我们可以利用所有可用的硬件来解决DW查询面临的挑战。

You will notice they are verydifferent in terms of goals but have a common discipline SQL. The SQL needs to be appropriate, well optimized etc. In reality 95% of Real World Performance Engineers time is spent on SQL design, execution and optimization. This is where the most leverage to increase performance exists.

你会注意到这两种环境在优化目标上差别很大，但有一个共同点就是SQL。 SQL都需要被优化。在现实中，95％的真实世界性能工程师的时间花在SQL设计，执行和优化。 这是提高性能的关键所在。

Q6：During your 27 years of system performance tuning, there got besomething that is unforgettable, would you like to share some little storieswhich impressed you the most？

问题六：在您27年的性能优化的生涯中，想必有一些难忘的经历吧？有没有什么让您印象最深刻的事情呢，跟大家分享一下吧？

Andrew ：I guess my first performance challenge which was an implicit data typeconversion problem is the most memorable for many reasons:

- It is possible to speed up systems by orders of magnitude !

- People are still making the same mistake today !

- I suspect they will be making the same mistake years after I have sailedaway from the database world!

我想我在性能优化中第一个挑战是一个隐式数据类型转换问题，是最令人难忘的。原因有很多：

- 可以加快系统的数量级！

- 人们今天仍然犯同样的错误！

- 我怀疑他们将在数据库世界航行之后，犯同样的错误几年！

Q7：People called the RWP, ‘A Genius Team’，you have made difference in the database especially the performance tuning area, meanwhile，you are giving speech, sharing new study all over the world, it’s so much more than just technology，so can you tell me what is the mission or belief of RWP, what makes you keep fighting and pursuit excellence?

问题七：在外界，RWP被称为天才团队，你们的团队的确给行业带来了很多变化和巨大的影响，除了技术本身的提高，我了解到你们也通过各种方式在很多地方进行技术分享和交流，您可以简单谈一下你们团队的使命是什么吗？是什么样的信念让你们坚持钻研也坚持分享？

Andrew ：The goal is to demonstrate and educate our customers the true power andvalue of the the Oracle database technologies. I am very lucky I have avery talented team who share my passion for the subject and are relentless to find the next performance gain even after they made a system 1000 times faster,why not go for 10,000 times faster !

我们团队的目标是展示和分享Oracle数据库技术的真正力量和价值给我们的用户。 我很幸运，我有一个非常有才华的团队，他们能够跟我一起探索和进步，实现我的想法。当我们使性能提升1000倍的时候，我们会无情寻找下一个性能增益，为什么不去快一万倍！

Q8：With the development of the Big data and cloud computing, Mobile Internet scenes are being imported into traditional industries，which results in the integration and rectification of IT system. Does this phenomenon bring in any performance issues to IT system, and how to avoidor handle them?

问题八：随着大数据和云时代的到来，移动互联的场景导入到传统行业中，这使得很多企业都面临着IT系统的变革，云化是普遍的趋势，这种现象会给IT系统带来哪些新的性能问题，如何正确应对呢？

Andrew ：Many of these technologies are very real and fashionable at the same time. The challenge is to identify the best bits of new technologies andlearn to exploit them for an advantage and not deploy them because they are seen a cool and new. In reality 1st generation systems with new technologies usually fail to meet their expectations because the technology isso new and the operators are not fully aware of how to fully exploit them.

这些技术非常真实，也成为大趋势。 面临的挑战是确定最佳的新技术，并学习利用它们的优势，而不是部署它们。在现实中，具有新技术的第一代系统通常不能满足他们的期望，因为技术是新的，并且运营商不完全知道如何充分利用它们。

The 2nd generation of systems that learn from the mistakes of the 1stgeneration always yields a better system. The challenge we face is by thetime we reach a second generation there is a brand new technology that looks glamorous and, surprise, we see another 1st generation project that will make anew set of mistakes.

第二代系统从第一代的错误中学习总是产生一个更好的系统。 我们面临的挑战是，当我们到达第二代时，有一种全新的技术，会更加完美，容易被接受。但很多人并没有在这样的系统升级和更新中吸取更多的经验，当再次出现一个新的系统和技术的时候，还是面临同样的“第一代”问题。

In summary we face many dilemmas when to inject new technologies into aproject. Again many of the problems are human, political and perception basedand have nothing to do with writing good applications.

总之，我们在关于何时将新技术注入项目上总是很谨慎， 再次，许多问题是人，政治和感知为基础，与编写好的应用程序无关。

Q9：The oracle optimizer changes with the database version updating, so would you like to share some significant changes happens to oracle optimizer inoracle database 12c release 2, which you most concerned about, and how wouldthese changes influence the performance of database system?

问题九：Oracle数据库的优化器在版本的更新中不断改进，在12.2中有哪些关于优化器的改进是您比较关注的呢，或者您觉得比较重要的一些改变,可以跟大家分享一下

Andrew ：The Oracle Optimizer in 12.1 has had a difficult introduction. We incorporated a number of adaptive features into the 12.1 optimizer that have not been well understood by DBAs at upgrade time. As a result of feedbackfrom the customers, in later versions of 12.1 and 12.2 we have turned a few of these features off by default to smooth the upgrade process from 11.2 to 12.1to 12.2 . The Real World performance team has a recipe for this process which has been very successful around the world. I would recommend people follow this recipe very carefully.

12.1的优化器中有一些改进是不合理的。 我们在引入了许多自适应功能，在升级时DBA不能很好地理解和使用。 作为客户反馈的结果，在12.1和12.2的更高版本中，我们默认关闭了一些这些功能，以使升级过程从11.2平滑到12.1到12.2。RWP团队在进行性能优化的时候回有明确的原则和制度的，不断尝试和挑战，不怕失败，这才让我们能够真正站在真实的环境的考虑下做优化。

Q10：Do you have some suggestions for the database operation, about how to detect、recognize , prevent, process and optimize the performance issue?

问题十：您有什么建议给运维者的吗，关于如何监测问题，识别问题、防范、处理和优化数据库遇到的各种性能问题？

Andrew ：The key to making performance gains is to look for leverage to effect change of the performance characteristics. Unfortunately too many DBAs spend time tuning platform specific features rather than looking for algorithmic enhancements to the applications. Platform tuning is only likely to yield incremental gains rather than order of magnitude gains.

提高性能的关键是寻找影响性能特性变化的关键点。 不幸的是，太多的DBA花费时间调整平台特定的功能，而不是寻找应用程序的算法增强。 平台调整只能产生同一级别上是性能增益，而不会有根本性的改变。

We describe platform tuning as bottom up performance work where as Real World Performance techniques are top down or a holistic approach toperformance.

我们将平台调优描述为自下而上的性能工作，而RWP的性能技术是自上而下的或整体的方法来执行。

Q11：For all the Oracle database technology enthusiasts, some of them might watching this right now, we’d like to know more about RWP team, like how many people do you have, how often do you give speech in every continent you have been cooperation with, how do you manage your team, and something like that, if you would like to share.

问题十一：有很多的技术爱好者可能会关注这次的采访，我们很想了解关于RWP更多的信息，关于你们的团队呀，管理呀，或者分享计划之类的东西，您愿意跟大家介绍一下吗？

Andrew ：The Real World Performance Teamis truly a global team. Somewhere in the world at any point in time there is RWP engineer working on a problem. Likewise in any month there will be RWP training events in every continent taking place. A few months ago we had classes in Europe, USA and China all running at the same time.

RWP团队是一个真正的全球性团队。 几乎在世界的某个地方在任何时间点都有RWP工程师在处理问题。 同时，我们每个月都会在不同大陆举行RWP培训活动。 就在几个月前，我们在欧洲，美国和中国都有同时进行的课程。

The team is very small team and this helps keep the quality of the solutions to a high level. We are lucky we can be quite selective on the performance projects we engage in. We tend to keep away from projects that really have little motivation toachieve big performance gains.

RWP团队是非常小的团队，这有助于保持高水平的解决方案的质量。 我们很幸运，我们对要参与的绩效项目有很大的选择性。我们倾向于远离那些真正没有动力去实现巨大业绩增长的项目。

I think the most important thing about the team is our ability to share ideas no matter how crazy they may seem, laugh and have fun in our work. I have learnt if people are being challenged and having fun they tend to grow professionally and produce greatresults.

我认为团队最重要的是我们分享想法的能力，无论这些想法看起来多么疯狂，我们都会乐在其中。 如果人们能享受挑战所带来的乐趣，他们的专业技能增长会更快，并且能够在专注的方向上取得成就。

3月28，Andrew大师与你相约北京，约吗？

活动详情及报名方式：

加入"云和恩墨大讲堂"微信群，参与讨论学习

搜索 盖国强（Eygle）微信号：eyygle，或者扫描下面二维码，备注：云和恩墨大讲堂，即可入群。每周与千人共享免费技术分享，与讲师在线讨论。

关注公众号，获得后续精彩分享返回搜狐，查看更多