A2 初級 美國腔 1108 分類 收藏
Back when I was taking up my current role, Minister of Science and Technology
a number of journalists asked me:
"Mr.Chang, are you convinced that cloud computing will be successful in Taiwan?"
To be honest, the answer was no.
It wasn't that we had no hope for success at all,
but it was definitely very challenging.
Why is that so?
Actually Ben already touched on this topic.
Changing minds really isn't easy, right?
So many preconceptions.
So what I want to report today are the challenges that we encountered,
and what are the preconceptions that we need to change.
The first question:
Is it safe to store our data on the cloud?
Let's put the Cloud aside for a while.
Let's first talk about 50, 100 years ago, when banks first opened.
Our old folks who lived through that time
were taking money out from under the bed and depositing them into the banks.
They were faced with a similar question.
After putting money into the banks, some folks even commented on the fact that
the bills which they drew from the bank
weren't the exact same notes that they deposited!
We face the same problem today
Today, I can put a computer into a data center.
But when I'm accessing the data center, I may not be using the same computer.
Some may feel that they lost out because they don't get to use their own computer.
But when you deposit an old bill into the bank
and get a new bill in return, you wouldn't think of that as "losing out"
So let's come back to present day.
Today, people who have lots of gold and jewellery
wants to deposit them at the bank. Why?
Because the bank promises to give you a safer environment than your home, right?
It's the same thing today.
Today, when we put our data onto the cloud,
the operators have to be able to give you a safer environment than your home.
Then you might ask me: I can control everything at my own company
and make sure that my data is secure.
Can a cloud operator do the same?
Like Ben said just now, if he couldn't
then he wouldn't be doing this for a living. So of course he can.
Second myth: the utilities at cloud data centers are very expensive
The electric bill at Acer's data center is 400 million Taiwanese Dollars every year.
Our basement water tank holds 6000 tons of water.
Some people say that this is not environmentally friendly.
But let me ask you.
Why do we encourage people to take buses nowadays?
How far can a bus run on a liter of fuel?
2 kilometers if you're lucky
With your own hybrid car you can probably go 10km on that same liter of fuel.
Then why are we encouraging you to take the bus instead of driving your car?
Because the bus has the economy of scale.
Likewise, today we are encouraging people to put their computers into data centers
and through the economy of scale in the data center
we can reduce the average amount of power and water used up by individual computers.
But some people like my previous boss, Chairman J.T. Wang says:
"But then all that carbon footprint becomes Acer's! That's unacceptable!"
It's true that we don't yet have a standardized way
of calculating carbon footprint.
But of course Acer alone can't bear the entire carbon footprint.
So who should bear it?
The carbon footprint should be apportioned
to the users of the data center based on how much electricity they use.
Similarly for Google's data center in Chang Hua County, which you mentioned.
That's the same thing.
Google's annual electric bill will be at least 2 billion for sure.
So whose carbon footprint is it? Google's? Or Taiwan's?
Google may be using this data center to serve other Asian countries
like Japan or Korea.
The carbon footprint should go to them, not Taiwan.
Taiwan is doing it's part for the environment by running this data center.
What about challenges?
We talked about software.
What's wrong with software?
If you read the papers, you'd know that
when we were selling tickets for A-Mei's concert,
the computer hanged before we could finish selling the tickets.
A few years ago, we were selling train tickets for the eve of Chinese New Year.
But the server hanged before we could finish selling.
These are not hardware problems. They are software issues.
We see a lot of apps being written by almost everyone nowadays and so we ask:
If you could write handphone apps, then why not software?
Let me ask you, when you're running Angry Birds on your handphone,
how many people are using that software? Just you, right?
But when we are selling train tickets on the eve of Chinese New year
how many people are trying to buy those tickets?
At least thousands if not tens of thousands.
A software meant for thousands of simultaneous users
is very different from a software written for just one user.
Does Taiwan have this capability then?
That's a good question. No one knows the answer.
So should Taiwan do it?
Let's talk about Google's Gmail first.
Is email software programming easy?
It's a very basic function and writing it seems to be pretty easy.
But if you see how many people in Google are working on Gmail
you wouldn't want to try and create a competing email software.
You and I may have never been to Gmail.
The Gmail department by itself pretty much takes up an entire building.
So this tells us that developing software isn't so easy.
Most importantly, transactional behavior is a problem without a technical solution.
What is transactional behavior?
Let's go back to the train tickets. On the eve of Chinese New Year
the most popular Taroko Express has 8 cars, 400 seats.
Ten thousand people are fighting for these 400 tickets.
Every seat was snatched up in an instant.
The data in the database is changing constantly.
At the same time you want me to make 10, 20 copies of the data for multiple users?
Sorry, can't be done.
Within an instant all the seats are sold out.
The database must be updated immediately so the same ticket isn't sold twice.
How can you, at the same time, make so many copies of the data?
This technical bottleneck has no solution in the short term.
So it's actually not that simple, not simple at all.
Lastly, where is Taiwan's opportunity?
We know that the Cloud has infrastructure, platform and application software.
Each of these links on the value chain has different characteristics.
You can sell a lot of servers so your main aim is to increase volume.
But when you talk about software, that's a high value product.
For example our healthcare system. How many copies can you sell?
Just one.
Because only Taiwan is using this software.
The healthcare of mainland China and USA are completely different.
We are switching to our second generation healthcare system soon.
You think you can sell a second copy of this system?
No way. There won't be a second sale.
But is this healthcare software valuable?
The whole of Taiwan is relying on this software when we go to the doctor's.
The value is very obvious. But what about volume?
I don't know which company the project was awarded to
or how much they were paid, but there's no volume to speak of.
On the other hand, we were speaking of servers.
We've sold hundreds of thousands, perhaps even millions of servers around the world.
But there's basically no difference between brand A and brand B.
That's why we put the emphasis on production volume.
As for developing application softwares
many people ask me: Does Taiwan stand a chance?
Maybe not.
We must develop software to benefit those who use healthcare
to help those who take electronic receipts.
At this time, I'd like to quote Deng Xiao Ping
"Be it a black cat or a white cat, a cat that can catch mice is a good cat".
Whether a product is domestic or imported doesn't matter
as long as it gives us good healthcare
and allows us to see the doctor with no fuss, then it is valuable to us.
It doesn't matter who made the software,
first we must ensure that we deliver on welfare
and then one day when we figure out how the imported software is made
we can clone it domestically and replace it.
But our final goal is to implement Cloud technology
so that when you have to visit the doctor
your visit is a smooth one, with no data issues.
When you buy your train tickets it won't matter how popular the date is
you can still buy your ticket within a second.
Instead of always seeing messages like
"Sorry, the server is busy right now, please try again in five minutes."
Let's hope never to see such messages again.
Thank you. (Applause)


【TEDx】台灣雲端計算的迷思與挑戰 (Simon Chang at TEDxTaipeiChange 2012)

1108 分類 收藏
Amy.Lin 發佈於 2017 年 1 月 5 日
  1. 1. 單字查詢


  2. 2. 單句重複播放


  3. 3. 使用快速鍵


  4. 4. 關閉語言字幕


  5. 5. 內嵌播放器


  6. 6. 展開播放器


  1. 英文聽力測驗


  1. 點擊展開筆記本讓你看的更舒服

  1. UrbanDictionary 俚語字典整合查詢。一般字典查詢不到你滿意的解譯,不妨使用「俚語字典」,或許會讓你有滿意的答案喔