Sunday, October 31, 2021

Stock buy back: XOM, CVX, VALE, RDS, FB, UBS, PG, AA, WFC, KDP, ASML, CMCSA, FANG, MSFT, CI, M, TGT, EBAY, UPS, GS, NVR, BP

 Dear:


The chatting history for this WeChat group "投资小群" is shown below.


————— 2021-10-31 —————

陈建敏 21:15

【美股派息回购一览】持续更新中

2021.10.29 埃克森美孚石油(XOM.N):将从2022年起启动高达100亿美元的回购计划。 回购股票的速度取决于市场状况。

2021.10.29 雪佛龙(CVX.N):宣布回购6.25亿美元的股票。将在第四季度增加股份回购。预计第四季度股票回购计划将处于指导区间的高端。

2021.10.29 淡水河谷(VALE.N):将在18个月内回购最多2亿股股份。

2021.10.28 荷兰皇家壳牌(RDS.A.N):本季度内已完成10亿美元的股票回购。

2021.10.26 Facebook(FB.O):将股票回购授权增加500亿美元。董事会授权回购500亿美元股票,并宣布将以两项业务的形式报告业绩,包括其核心的 Family of Apps 和 Facebook Reality Labs。

2021.10.26 瑞银(UBS.N):计划在第四季度回购至多6亿美元的股票。

2021.10.19 宝洁(PG.N):公司通过派发20亿美元的股息和回购近30亿美元的股票向股东返还近50亿美元现金。

2021.10.15 美国铝业(AA.N):将开启价值5亿美元的股份回购。将派发现金股息并开启新一轮股票回购计划。五年来首次派息。

2021.10.14 富国银行(WFC.N)CFO:可能进行回购超180亿美元的资本计划。通过53亿美元的回购增加了公司股份持有人的收益。

2021.10.13 高通宣布新一轮100亿美元的股票回购。

2021.10.8 截至9月底,淡水河谷(VALE.N)股票回购支出为48亿美元。

2021.10.1 美国饮料制造商Keurig Dr Pepper(KDP.O)据悉计划回购高达40亿美元的股票。

2021.10.1 荷兰国际银行(ING.N)宣布一项规模为17亿美元的股票回购计划。股票回购计划将于2021年10月5日开始,预计不迟于2022年5月5日结束。

2021.9.29 阿斯麦(ASML.O):预计将继续提高股息和扩大股票回购计划。

2021.9.24 麦当劳(MCD.N):恢复股票回购。

2021.9.23 戴尔(DELL.N)宣布批准高达50亿美元的回购计划。

2021.9.22 康卡斯特(CMCSA.O)CEO:回购和增加派息仍是首要任务。

2021.9.17 石油天然气公司Diamondback Energy Inc(FANG.O)表示将加快向股东返还资本并启动股票回购计划。

2021.9.15 微软(MSFT.O):授权至多600亿美元进行股票回购。

2021.8.24 Cigna(CI.N)宣布20亿美元加速股票回购协议。

2021.8.20 Spotify(SPOT.N)宣布至多10亿美元股票回购。

2021.8.19 梅西百货(M.N):授权一项规模为5亿美元的回购计划。

2021.8.18 塔吉特(TGT.N):董事会批准了一项新的规模为150亿美元的股票回购计划。

2021.8.12 eBay(EBAY.O):将股票回购价值扩大至30亿美元。

2021.8.6 联合包裹(UPS.N)宣布派发季度股息以及批准了一项新的规模为50亿美元的股票回购计划。

2021.8.5 高盛(GS.N)本年至今股票回购总额达6830亿美元,为历年同期第二大。

2021.8.5 住宅建设商NVR Inc.(NVR.N)宣布以5亿美元回购公司股票。

2021.8.3 英国石油(BP.N):将开始规模14亿美元的回购计划。

David Cote '76 UNH 2011 Commencement Speech

Oct. 31, 2021

Here is the link. 

David Cote, University of New Hampshire Class of 1976, was the Commencement Speaker at the 2011 Ceremony. His speech was named to NPR's list of Best Commencement Speeches, Ever: http://apps.npr.org/commencement/spee...

7:11/ 

What I have learned | Condensed to four major areas 

  • First Recognize the importance of people and your own behaviors
  • Be self-aware, be honest with yourself and face reality 
  • Know who you are in your personal characteristics, what are your strengths and what are the areas where you are not so strong. 
  • You can't be good at everything so how can you fill in where you aren't and I can tell you when you are younger this is one of the toughest things to do. 
  • We're all born with different skills and aptitudes, I was fortune to be born at a time when my skill with a pencil was valuable a hundred years ago
  • I'd have probably just been a bad farmer now , that is not something you can control
  • What you can control are your behaviors, how you look at life, how you maximize the skills you do have
  • Be aware of the impact that you have on people around you
  • Do people feel better as a result of spending time with you during the course of your career
  • Your biggest accomplishments will be achieved by motivating and inspiring others 
  • Develop personal credibility in your dealings 
  • Do what you say is not an empty phrase
  • Stand for something and have values 
  • Be open to new ideas
  • Have opinions and sate them but be willing to modify them for a better suggestion 
  • Acknowledge the contribution of others 
  • Even if it's painful it reflects on you recognize because someone disagrees you
  • Don't want to make it personal, at the same time not everybody you deal with is reasonable 
  • There truly are zealots in the world who can only see their point of view and these people are the most dangerous and they can be found in all walks of life
  • Be tolerant if a decision doesn't see right
  • Change it better the embarrassment of a changed mind, then the lifetime penance of a bad decision in business and in life
  • Be prepared to take hits to your ego whether deserved or not 
  • My dad was a very proud man who could have quite a temper, he owned a service station and one day when I was working with him when I was about 14, I watched him take an unwarranted verbal beating from a customer, I was shocked because three was no response when the customer left, he came back and sat next to me on the curb and said Dave, sometimes in business, sometimes in life you have to put your pride in your back pocket. Don't let anyone break you down emotionally. Be able to take that hit, they will happen. 
  • Second get out of your comfort zone, better yet get comfortable about being out of your comfort zone. 
  • Be a learner. The greatest learning occurs where you're uncomfortable because you don't know it all. 
  • Push yourself. Be willing to take a chance. Take a risk with something, new career, hobbies, dating, sports whatever.
  • It doesn't mean you should be impulsive. It doesn't mean you should take calculated risks. 
  • Get yourself into the arena. Be a participant. Not a spectator. Don't always wait for approval or perfect knowledge. 
  • Do something. 
  • Have the confidence in yourself to take that chance. 
  • If you don't certainly others won't and there will always be plenty of people to tell you why it won't work including some friends and family and sometimes they'll be right and that's painful.
  • And as my mom always said think for yourself. 
  • Just because everyone is doing something, it doesn't mean it's a right which also means you have to be able to handle rejection and failure. 
  • Successful people have failures, sometimes spectacularly embarrassing ones, but they learn from it and move on to the next success. 

Hon stock 200 share purchase -> Honeywell business study -> Honeywell CEO David Cote | Weekend study

 President, chief executive officer, and chairman, Honeywell International

Nationality: American.

Born: 1952.

Education: University of New Hampshire, BS, 1976.

Family: Married twice; children: two from first marriage, one from second marriage.

Career: General Electric, 1974–1976, factory laborer; 1976–1996, manager; 1996–1999, senior vice president, president and CEO of Appliances; TRW, 1999–2001, president and COO; 2001, president and CEO; 2001–2002, president, CEO, and chairman; Honeywell International, 2002, president and CEO; 2002–, president, CEO, and chairman.

Awards: Honorary Juris Doctor, Graziadio School of Business at Pepperdine University, 2001.

Address: Honeywell International, 101 Columbia Road, Morristown, New Jersey 07962; http://www.honeywell.com.

■ In a speech at Pepperdine University in April 2001, David M. Cote outlined four areas in which businesspeople should focus: recognizing the importance of people and one's own behaviors; getting out of one's comfort zone; having goals and being results oriented; and enjoying life. Cote's success in managing people may have stemmed from his understanding the needs of others. His shifting from tough job to tough job certainly reflected his continuous desire to extend beyond his comfort zone. Throughout his career, he developed a reputation for being very goal oriented. As for happiness: as a driven man, devoted to performance, he derived profound enjoyment in meeting and overcoming challenges.

GENERAL ELECTRIC

When he graduated from high school, Cote seemed to have no goals, belying the driven temperament that would eventually propel him to the apex of his profession. He decided to skip university, instead using his college money to buy a car, and worked as a manual laborer; after a couple of years he realized that he would not excel in that field. Thus, he entered the Wittemore School of Business at the University of New Hampshire. His studies were protracted by his full-time night job in a General Electric jet-engine manufacturing plant, as well as by a period during which he also purchased a boat with a friend and worked as a lobster fishermen. He graduated from college after approximately six years.

After receiving his bachelor's degree in business administration, Cote worked as a manager in the consumer-electronics, jet-engines, and plastics divisions at General Electric. The company was noted for its goal-oriented style of operations, and Cote readily absorbed its management doctrine. In 1996 the business world learned what upper management at General Electric seemed to already know—that Cote was an innovator who could aptly refine the company's business practices, cutting costs and improving sales. That year, he was named corporate senior vice president and also became the president and CEO of the $6 billion Appliances division.

Cote soon became the leader of General Electric's "smart bomb" technique for generating sales in Asia. Instead of creating a marketing plan for the entire continent, he formed bubble teams of sales representatives, engineers, and cultural experts to study each Asian nation individually, with the teams eventually creating bundles of appliances suited to the cultural groups of each individual nation. General Electric Appliances registered profits of about $100 million in each of the years Cote ran the division, and other General Electric divisions copied the "smart bomb" techniques. Cote also led one of General Electric's first joint-venture efforts, forming a partnership with Culligan to manufacture refrigerators with built-in water filtration. In what would be a hallmark of his management, Cote streamlined the appliances division by emphasizing the specific goals that needed to be achieved.

TRW

When it became evident that Cote would not be the one to replace the retiring CEO Jack Welch, Cote looked elsewhere for new challenges, eventually finding them at TRW. In November 1999 he became TRW's president and COO, with the understanding that he would replace the CEO Joseph Gormanin 2001. At TRW, Cote introduced the Six Sigma management system, a doctrine that focused the production process on reducing defects in products to less than 3.4 per million opportunities; "opportunities" were defined as occasions when a defect could occur, and "defects" were defined as any results outside of customer specifications. As a whole, the Six Sigma system made the manufacturing process especially responsive to customer needs.

In February 2001 Cote became CEO as well as president of TRW; in August 2001 he added chairman of the board to his title. Cote led the creation of the TRW subsidiary Velocium, which manufactured ultra-high-speed semiconductors and was just beginning to make itself felt in the marketplace when Cote departed from TRW.

HONEYWELL INTERNATIONAL

Honeywell International had earlier tried to merge with General Electric, only to have the merger rejected by antitrust authorities in the European Union; having reorganized to become part of General Electric, the company had trouble reorienting itself to remaining an independent entity. The former chairman Lawrence A. Bossidy was brought in to save Honeywell from collapse, and when he retired for a second time he selected Cote to be his replacement.

On February 19, 2002, Cote was named president, CEO, and a member of the board of directors for Honeywell. On July 1, 2002, Cote was elected chairman of the board after Bossidy retired. Cote had his hands full: although Honeywell grossed $24 million in 2001, the company faced mounting debts, settlements for lawsuits stemming from employees' exposure to asbestos, a stock that would fall 31 percent in 2002, and demoralized management. Cote knew that running the huge company, with 115,000 employees in 95 countries, would prove challenging. He quickly applied his goal-oriented management philosophy to Honeywell, making cash, growth, people, productivity, digitization, and Six Sigma the focuses of his administration.

Honeywell lost $220 million in 2002, partly because of asbestos-suit payouts, the purchase of the sensors business Invensys for $416 million, and slow sales. However, by selling other Honeywell units and cutting costs—partly by sending some American jobs abroad to Romania and Singapore, where labor costs were lower, and partly by reducing defects in production—by the end of 2003 Honeywell had amassed about $2 billion in cash reserves. Cote was heavily criticized in the press for the $32 million in compensation he received in 2002, although he explained that the amount was intended to cover options that he had lost at TRW as a result of moving to Honeywell.

See also entries on TRW Inc. and Honeywell Inc. in International Directory of Company Histories .

sources for further information

Barrett, Amy, "In the Credibility Penalty Box: Can Honeywell CEO Cote Restore Investors' Confidence?" BusinessWeek , April 28, 2003, p. 80.

Cote, David M., "Keynote Commencement Address: The Graziadio School of Business and Management," Pepperdine University, April 14, 2001, http://www.pepperdine.edu/PR/NotableSpeeches/cote.htm .

Grant, Linda, "GE's Smart Bomb Strategy," Fortune.com , July 21, 1997, http://www.fortune.com/fortune/subs/article/0,15114,380002,00.html .

Murphy, Tara, "Honeywell's New CEO Is the Right Man," Forbes.com , March 5, 2002, http://www.forbes.com/2002/03/05/0305bigcap.html .

—Kirk H. Beetz



Read more: https://www.referenceforbusiness.com/biography/A-E/Cote-David-M-1952.html#ixzz7ArKlKrzD

Deming Cup Winners Lecture: David Cote

Oct. 31, 2021

David Cote, Chairman and CEO, Honeywell, delivered his Deming Cup Winners Lecture on February 6, 2017.

Here is the link. 

Leadership training | David Cote | How to be a leader? 

First 25 minutes presentation by David Cote 

  • Get all facts, all opinions and make decisions. 
  • Be prepare to make decisions. 
  • Be the right at the end of the meeting. 
  • Do not know your position - Do not say anything 
  • Encourage discussion - make good decision 
  • How to get people to participate? 
  • Leadership classes - 25 year old, appraisal - 39, 40 years old defensive? 
  • Every single meeting - who, what and when? 
  • Be confident on your decision - see if it works - not sure - you have to be sure ...... Second guessing - input points 
  • During second quitting - business acumen - What is your biggest learning? 
  • Business world, result not effort, not hourly worker 
  • Push people to think about 
  • Get long term - superball - come back, setbacks, commitment to change, for example, honeywell user experience - use our product and services - 
  • Being smart is not good enough, good judgement is important, it takes more than that. 
Questions asked by audience:

  1. Lean organization, for example, Toyota production system - Learn Toyota production system two weeks, buy concept, 1 hire and 100 interviewees; 
  2. Bring job back to USA? How to analyze? 
  3. Career path - major company CEO
  4. Get six years to finish college - I could not support myself, how can I support a kid? Hourly employee, and got lucky as an employee to have hourly pay, whatever job to pay me more. I just need more money, 5 years, two kids, divorce 25 years, what I do right now? 60 days a year. Interesting to learn what I like the business, in GE, 25 years old, often recommendation -> general manager -> HR from finance to general manager, rather than a general manager of small team, 3 people -> CEO somehwere -> Start to think about that way, 39 or 40 years old 
  5. Continue on item 4, path to have a career -> how do I figure out? As fast as I could. 
  6. New question  about culture - 




Saturday, October 30, 2021

David Cote, Former Chairman and CEO of Honeywell

Oct. 30, 2021

Here is the link. 

In this episode of Talks at GS, David Cote discusses lessons from his time at the helm at Honeywell, as described in his new book "Winning Now, Winning Later," along with his views on leading through crisis and the future of digital transformation. Learn More https://www.goldmansachs.com/insights...

HON stock purchase -> Honeywell CEO -> David M. Cote | My 20 minutes study

 David M. Cote (born July 19, 1952[2]) is an American businessman. Cote previously worked for General Electric[3] and TRW Inc.[4] before he was appointed chairman and chief executive officer (CEO) of Honeywell in 2002,[4] following their acquisition by AlliedSignal. Cote also sat on the JP Morgan Chase risk committee during the period in which the firm lost $6 billion trading credit derivatives.[5] Cote stepped down as CEO at Honeywell at the end of March 2017 and was succeeded by Darius Adamczyk.

Early life[edit]

Cote was born in Manchester, New Hampshire and graduated from Pembroke Academy in Pembroke, New Hampshire in 1970.[6] The following year, Cote enrolled at the University of New Hampshire,[3] and while attending UNH full-time, he worked an hourly job on the night shift at a nearby GE jet engine plant.[3] In 1976, he graduated from the University of New Hampshire with a bachelor's degree in business administration.[3][7]


Career[edit]

General Electric[edit]

Cote joined General Electric full-time in November 1976, and worked there for more than twenty years. He transitioned from his hourly production work at the General Electric aircraft-engine plant in New Hampshire to a full-time auditing job at another GE plant in Massachusetts.[3] In 1985, his handling of an interaction with CEO Jack Welch became the catalyst for Cote's advancement at GE.[3] Welch promoted Cote three levels of management.[3] Over his career there, he held positions in manufacturing, finance, marketing, strategic planning and general management,[7] before becoming CEO of GE Appliances in 1996.[7]


TRW[edit]

In November 1999, Cote joined TRW as president and COO where he introduced the Six Sigma management system to reduce defects in manufacturing.[4][8] In February 2001, Cote was appointed CEO, and later chairman of the board.[9] Cote led the creation of the TRW subsidiary Velocium, which manufactured ultra-high-speed semiconductors.[4]

In February 2002, Cote announced he would be leaving TRW. Reportedly, the news surprised employees and some executives, who learned of Cote's departure hours before the announcement.[10][11]


Honeywell[edit]

Honeywell selected Cote as successor to Lawrence Bossidy,[12] following the AlliedSignal acquisition of Honeywell[3] and European Union's rejection of Honeywell's merger with General Electric.[12] Cote was elected CEO, president, and a member of the board of directors on February 19, 2002. He was elected chairman of the board on July 1, 2002.[12]

The year Cote took office, Honeywell lost $220 million.[7] Cote instituted conservative accounting to streamline costs. In an effort to reduce the unpredictability of asbestos-plus-environmental expenses, Honeywell established a trust for claims and reclaim soil at chemical plants. As a result, that expense is consistently $150 million a year, after-tax.[3] Honeywell saw improved quality in design, increased production, and lower production costs after it implemented a new productivity management system.[13][14] During the 2008-2009 recession, the company instituted furloughs to reduce overall operating costs rather than lay off workers.[3][15] During his tenure at Honeywell, the company's donations via its Political Action Committee rose from rose from two hundred thousand dollars in 2002 to 8 million in 2014.[16] As of June 2016, Honeywell's market value has risen from $28 billion to $87 billion since 2002.[17][18]

While CEO of Honeywell International in 2015, Cote earned a total compensation of US$25,053,000 which included a base salary of $1,890,000, an annual bonus of $5,700,000, and $10,338,000 in stock options.[19]

Cote stepped down as CEO at Honeywell at the end of March 2017 and was succeeded by Darius Adamczyk. Cote continued as executive chairman through April 2018.[20][21][22]


Appointments[edit]

Banking[edit]

Cote was a member of the board of directors at JPMorgan Chase and an advisor to Kohlberg Kravis Roberts (KKR). In 2012, Cote came under criticism as one of the three members of JP Morgan Chase's risk committee, after CEO Jamie Dimon said on May 10, 2012 that the firm's chief investment office suffered a $2 billion loss trading credit derivatives.[23] Commentators identified a lack of relevant expertise among the members of the committee, identifying Cote and a museum official who also served, in particular for their lack of banking experience.[24]

In February 2014 it was announced that Cote would fill a vacancy on the board of the New York Federal Reserve. Simon Johnson, former chief economist of the International Monetary Fund, writing in The New York Times, raised doubts about the suitability of Cote's appointment, noting the "systematic breakdown of compliance and risk control" during the period when Cote was on the board of JPMorgan Chase, whilst also noting that some, but not all, of the problems there pre-dated Cote's appointment.[25] The election papers did not mention Cote's role at JPMorgan Chase in his candidate C.V.[26] Cote stepped down from the board of New York Federal Reserve in March 2018.[27]


Other[edit]

Cote was a member of the Executive Committee of The Business Council for 2011 and 2012.[28]


Books[edit]

Awards[edit]

  • 2013-2016 World's Best CEOs, Barron's[38][39]
  • 2013 Chief Executive of the Year, Chief Executive Magazine[40]


Honeywell CEO: 3 Steps to Success | Mad Money | CNBC

 Oct. 30, 2021

Here is the link. 

During his tenure at Honeywell, Dave Cote has certainly sweetened the pot. He sat down with Jim Cramer on Thursday in his last interview as the man sitting at the helm of the company.

Honeywell CEO Adamczyk on Sales, Growth, M&A

Oct. 30, 2021

Here is the link. 

Apr.22 -- Darius Adamczyk, chief executive officer and chairman at Honeywell, discusses the company’s first-quarter sales growth, global growth opportunities, and taking a disciplined approach to investments and M&A. He speaks with Bloomberg's Vonnie Quinn on "Bloomberg Markets."


Honeywell International CEO: The New Honeywell | Mad Money | CNBC

Oct. 30, 2021

Here is the link. 

The Honeywell breakup is rapidly approaching and it could be a big win for shareholders. Jim Cramer checks in with spin-off overseer and Honeywell CEO Darius Adamczyk for the details.

Honeywell CEO on Return to Office, Travel, Delivery Tech

Oct. 30, 2021

Here is the link. 

Aug.04 -- Honeywell Chairman and CEO Darius Adamczyk discusses the outlook for returning workers to the company’s offices, global air travel recovery, and investments in delivery technology on "Bloomberg Markets: European Close."

Friday, October 29, 2021

Public School Character Development: Accountability, Conviction, What is Really Important?

Oct. 29, 2021

Here is the link. 

When Coach Carter learns that players on his team have not kept their grades up as per their contract, he locks the gym shut until they focus on learning.

Coach Carter Documentary-the real coach carter (2 of 2)

Oct. 29, 2021

Here is the link. 

A behind the scene and exclusive interview with Coach Ken Carter and The members of the Richmond Oilers

WHAT EVER HAPPENED TO THE RICHMOND 1999 TEAM ?

Oct. 29, 2021

Here is the link. 


Coach Carter Respect & Responsibility Theme

Oct. 29, 2021

Here is the link. 


Coach Carter Advocate Theme

Oct. 29, 2021

Here is the link. 



Coach carter clips | My notes

Oct. 29, 2021

Here is the link. 

Scene 2: Library 

3:42 / 7:21 - 5:15 

Scene 2: Library 

Get to the college, and get a better life | Percentage to go to prison - Richmond community

Richmond high - 50% graduate, 50% go to college after graduation 

20% get arrested. One of you will get arrested. 

Those are the numbers. Those are stats ... 

I see that a system will let you fail. Only one student will go to college. Probably go to prison. Get arrested, look at left, and look at right, 80 percent more go to prison then to the college. 



Healthcare stocks: XLV, TMO, UNH, ISRG

 Healthcare was the strongest sector. $XLV leaped 0.97% thanks to strength by $TMO$UNH, and $ISRG. Real estate fell 1.2%, energy slipped 0.66%, and utilities powered down 0.59%.

HTTP persistent connection | My 20 minutes reading time - Oct. 29, 2021 | Understand GFS paper

HTTP persistent connection

From Wikipedia, the free encyclopedia

 HTTP persistent connection, also called HTTP keep-alive, or HTTP connection reuse, is the idea of using a single TCP connection to send and receive multiple HTTP requests/responses, as opposed to opening a new connection for every single request/response pair. The newer HTTP/2 protocol uses the same idea and takes it further to allow multiple concurrent requests/responses to be multiplexed over a single connection.

Advantages[edit]

According to RFC 7230, section 6.4, "a client ought to limit the number of simultaneous open connections that it maintains to a given server". The previous version of the HTTP/1.1 specification stated specific maximum values but in the words of RFC 7230 "this was found to be impractical for many applications... instead... be conservative when opening multiple connections". These guidelines are intended to improve HTTP response times and avoid congestion. If HTTP pipelining is correctly implemented, there is no performance benefit to be gained from additional connections, while additional connections may cause issues with congestion.[13]

Disadvantages[edit]

If the client does not close the connection when all of the data it needs has been received, the resources needed to keep the connection open on the server will be unavailable for other clients. How much this affects the server's availability and how long the resources are unavailable depend on the server's architecture and configuration.

Also a race condition can occur where the client sends a request to the server at the same time that the server closes the TCP connection.[14] A server should send a 408 Request Timeout status code to the client immediately before closing the connection. When a client receives the 408 status code, after having sent the request, it may open a new connection to the server and re-send the request.[15] Not all clients will re-send the request, and many that do will only do so if the request has an idempotent HTTP method.

Use in web browsers

All modern web browsers including Google ChromeFirefoxInternet Explorer (since 4.01), Opera (since 4.0)[16] and Safari use persistent connections.

By default, Internet Explorer versions 6 and 7 use two persistent connections while version 8 uses six.[17] Persistent connections time out after 60 seconds of inactivity which is changeable via the Windows Registry.[18]

In Firefox, the number of simultaneous connections can be customized (per-server, per-proxy, total). Persistent connections time out after 115 seconds (1.92 minutes) of inactivity which is changeable via the configuration.[

The Google File System | Paper reading | Another two hours at least | Day one to read

 Oct. 29, 2021

Here is the link. 


2.3 Architecture 

A GFS cluster consists of a single master and multiple chunkservers and is accessed by multiple clients, as shown in Figure 1. Each of these is typically a commodity Linux machine running a user-level server process. It is easy to run both a chunkserver and a client on the same machine, as long as machine resources permit and the lower reliability caused by running possibly flaky application code is acceptable. Files are divided into fixed-size chunks. Each chunk is identified by an immutable and globally unique 64 bit chunk handle assigned by the master at the time of chunk creation. Chunkservers store chunks on local disks as Linux files and read or write chunk data specified by a chunk handle and byte range. For reliability, each chunk is replicated on multiple chunkservers. By default, we store three replicas, though users can designate different replication levels for different regions of the file namespace.

My notes:

  • Each chunk - 64 bit chunk handle - master assigned at the time of chunk creation 
  • Each chunk is replicated on multiple chunkservers. 
  • By default, we store three replicas - replication levels ? 
The master maintains all file system metadata. This includes the namespace, access control information, the mapping from files to chunks, and the current locations of chunks. It also controls system-wide activities such as chunk lease management, garbage collection of orphaned chunks, and chunk migration between chunkservers. The master periodically communicates with each chunkserver in HeartBeat messages to give it instructions and collect its state.

My notes:
  • HeartBeat messages - master <-> chunkserver 
  • master maintains all file system metadata - namespace, access control information, the mapping from files to chunks, and the current locations of chunks. 
  • Also system-wide activities such as chunk lease management, garbage collection of orphaned chunks, and chunk migration between chunkservers. 
GFS client code linked into each application implements the file system API and communicates with the master and chunkservers to read or write data on behalf of the application. Clients interact with the master for metadata operations, but all data-bearing communication goes directly to the chunkservers. We do not provide the POSIX API and therefore need not hook into the Linux vnode layer.

Neither the client nor the chunkserver caches file data. Client caches offer little benefit because most applications stream through huge files or have working sets too large to be cached. Not having them simplifies the client and the overall system by eliminating cache coherence issues. (Clients do cache metadata, however.) Chunkservers need not cache file data because chunks are stored as local files and so Linux’s buffer cache already keeps frequently accessed data in memory.

2.4 Single Master 

Having a single master vastly simplifies our design and enables the master to make sophisticated chunk placement and replication decisions using global knowledge. However, we must minimize its involvement in reads and writes so that it does not become a bottleneck. Clients never read and write file data through the master. Instead, a client asks the master which chunkservers it should contact. It caches this information for a limited time and interacts with the chunkservers directly for many subsequent operations. 

Let us explain the interactions for a simple read with reference to Figure 1. First, using the fixed chunk size, the client translates the file name and byte offset specified by the application into a chunk index within the file. Then, it sends the master a request containing the file name and chunk index. The master replies with the corresponding chunk handle and locations of the replicas. The client caches this information using the file name and chunk index as the key.

The client then sends a request to one of the replicas, most likely the closest one. The request specifies the chunk handle and a byte range within that chunk. Further reads of the same chunk require no more client-master interaction until the cached information expires or the file is reopened. In fact, the client typically asks for multiple chunks in the same request and the master can also include the information for chunks immediately following those requested. This extra information sidesteps several future client-master interactions at practically no extra cost.

2.5 Chunk Size 

Chunk size is one of the key design parameters. We have chosen 64 MB, which is much larger than typical file system block sizes. Each chunk replica is stored as a plain Linux file on a chunkserver and is extended only as needed. Lazy space allocation avoids wasting space due to internal fragmentation, perhaps the greatest objection against such a large chunk size.

A large chunk size offers several important advantages. First, it reduces clients’ need to interact with the master because reads and writes on the same chunk require only one initial request to the master for chunk location information. The reduction is especially significant for our workloads because applications mostly read and write large files sequentially. Even for small random reads, the client can comfortably cache all the chunk location information for a multi-TB working set. Second, since on a large chunk, a client is more likely to perform many operations on a given chunk, it can reduce network overhead by keeping a persistent TCP connection to the chunkserver over an extended period of time. Third, it reduces the size of the metadata stored on the master. This allows us to keep the metadata in memory, which in turn brings other advantages that we will discuss in Section 2.6.1.

My notes:

A large chunk size offers several important advantages:
  1. First, it reduces clients’ need to interact with the master because reads and writes on the same chunk require only one initial request to the master for chunk location information.
  2. The reduction is especially significant for our workloads because applications mostly read and write large files sequentially. 
  3. Even for small random reads, the client can comfortably cache all the chunk location information for a multi-TB working set.
  4. Understand the first point in step 1 to step 3. More details are from different angles in step 2 and step 3. 
  5. Second, since on a large chunk, a client is more likely to perform many operations on a given chunk, it can reduce network overhead by keeping a persistent TCP connection to the chunkserver over an extended period of time.
  6. Try to understand second point, a persistent TCP connection to the chunkserver over an extended period of time. What is persistent TCP connection? 
  7. Third, it reduces the size of the metadata stored on the master. This allows us to keep the metadata in memory, which in turn brings other advantages that we will discuss in Section 2.6.1.
On the other hand, a large chunk size, even with lazy space allocation, has its disadvantages. A small file consists of a small number of chunks, perhaps just one. The chunkservers storing those chunks may become hot spots if many clients are accessing the same file. In practice, hot spots have not been a major issue because our applications mostly read large multi-chunk files sequentially.

However, hot spots did develop when GFS was first used by a batch-queue system: an executable was written to GFS as a single-chunk file and then started on hundreds of machines at the same time. The few chunkservers storing this executable were overloaded by hundreds of simultaneous requests. We fixed this problem by storing such executables with a higher replication factor and by making the batchqueue system stagger application start times. A potential long-term solution is to allow clients to read data from other clients in such situations.

2.6 Metadata 

The master stores three major types of metadata: the file and chunk namespaces, the mapping from files to chunks, and the locations of each chunk’s replicas. All metadata is kept in the master’s memory. The first two types (namespaces and file-to-chunk mapping) are also kept persistent by logging mutations to an operation log stored on the master’s local disk and replicated on remote machines. Using a log allows us to update the master state simply, reliably, and without risking inconsistencies in the event of a master crash. The master does not store chunk location information persistently. Instead, it asks each chunkserver about its chunks at master startup and whenever a chunkserver joins the cluster.

2.6.1 In-Memory Data Structures 

Since metadata is stored in memory, master operations are fast. Furthermore, it is easy and efficient for the master to periodically scan through its entire state in the background. This periodic scanning is used to implement chunk garbage collection, re-replication in the presence of chunkserver failures, and chunk migration to balance load and disk space usage across chunkservers. Sections 4.3 and 4.4 will discuss these activities further.

One potential concern for this memory-only approach is that the number of chunks and hence the capacity of the whole system is limited by how much memory the master has. This is not a serious limitation in practice. The master maintains less than 64 bytes of metadata for each 64 MB chunk. Most chunks are full because most files contain many chunks, only the last of which may be partially filled. Similarly, the file namespace data typically requires less then 64 bytes per file because it stores file names compactly using prefix compression.

If necessary to support even larger file systems, the cost of adding extra memory to the master is a small price to pay for the simplicity, reliability, performance, and flexibility we gain by storing the metadata in memory.

2.6.2 Chunk Locations 

The master does not keep a persistent record of which chunkservers have a replica of a given chunk. It simply polls chunkservers for that information at startup. The master can keep itself up-to-date thereafter because it controls all chunk placement and monitors chunkserver status with regular HeartBeat messages.

Remember four things:
  1. a persistent record of which chunkservers have a replica of a given chunk
  2. the master polls chunkservers for step 1 information at startup
  3. the master monitors chunkserver status with regular HeartBeat messages 
  4. the master controls all chunk placement 
We initially attempted to keep chunk location information persistently at the master, but we decided that it was much simpler to request the data from chunkservers at startup, and periodically thereafter. This eliminated the problem of keeping the master and chunkservers in sync as chunkservers join and leave the cluster, change names, fail, restart, and so on. In a cluster with hundreds of servers, these events happen all too often.

Another way to understand this design decision is to realize that a chunkserver has the final word over what chunks it does or does not have on its own disks. There is no point in trying to maintain a consistent view of this information on the master because errors on a chunkserver may cause chunks to vanish spontaneously (e.g., a disk may go bad and be disabled) or an operator may rename a chunkserver.

MIT distributed system course recommended: Paper to read

Spanner: Google’s Globally-Distributed Database
https://pdos.csail.mit.edu/6.824/papers/spanner.pdf  


No compromises: distributed transactions with consistency, availability, and performance
https://pdos.csail.mit.edu/6.824/papers/farm-2015.pdf   

Resilient Distributed Datasets: A Fault-Tolerant Abstraction for In-Memory Cluster Computing
https://pdos.csail.mit.edu/6.824/papers/zaharia-spark.pdf  

Scaling Memcache at Facebook
https://pdos.csail.mit.edu/6.824/papers/memcache-fb.pdf  

Secure untrusted data repository
https://pdos.csail.mit.edu/6.824/papers/li-sundr.pdf

BitCoin
https://pdos.csail.mit.edu/6.824/papers/bitcoin.pdf

blockChain
https://pdos.csail.mit.edu/6.824/papers/blockstack-atc16.pdf