A note of HPC user community structure and "classic" vs "containerized/virtualized" HPC cluster dilemma

By Dr. Nikolai Bezroukov

It looks to me that putting all bets on the "Containerized" HPC cluster solution during upgrade of the Existing cluster HPC group actually puts the group between the rock and the hard place. There are two reasons for that. One is the success of "Classic" HPC cluster which pushed research forward and created somewhat exaggerated expectation as for what to expect from  the "Containerized" HPC cluster which might or might not materialize.  Add top this "oversell" from HOC cluster vendors. Of course "some dirt can always be swiped under the carpet" and deployment  of such a claster can be declared to be a huge success no matter what,  but  the problem does exist.

The  second reason in more complex and is based on the fact that the research community is not uniform. Simplifying we can distinguish between two important factions:

1. Silent majority. This faction is completely satisfied with "Classic" HPC cluster. Many of them are unable to use scheduler fully and have just promote borrowed from somebody submission scripts. Many use Quriosity only accidentally, say a couple of time a month or a quarter and thus has no possibility to acquire real experience with it, remaining "forever" on the basic level. Some use Quriosity via GUI which submits scripts and returns results and they do not use Quriosity directly. They are just aware about its existence and that this is "good thing". May be there are some other important subcategories.  While they might be able to use in some cases Docker containers for them Kubernetes and Azure are dirty words for them. They do not want all this additional complexity and overhead and they resent push into the cloud although they understand that to voice it would be politically incorrect.  For many the ideal is still to own a powerful workstation.
2. Vocal "progressive" minority. There are some alpha researchers in this group who really can use containers for their benefit and who are talented enough to understand the technology and its limitation,  but there are also "snake oil sellers" ( profiteers trying to exploit an unsuspecting public by selling it fake cures) -- people for whom the  "latest and greatest" technology is valuable not as technology, but as the mean to ensure their own survival and promotion (for example some advocates of "machine learning"). Small sub-faction of this faction can be called beta addicts and like any type of addicts they are dangerous and pushy people. Generally this faction tend to push the ball too far, and left unchecked can cause harm to other categories of researchers.

To a certain extent this structure of research community also is replicated internally within our own group: there  are people for who running  everything in container is the ultimate goal and there are people who view  the issue more realistically and ask important  question: "Containers or VM for which application?".

For quantum chemistry containerization this is not that impressive idea as they use MPI extensively and can benefit from direct access to Infiniband and less overhead from virtualized and split into groups of cores CPUs (they typically use all cores then can get on a particular computational node, so virtualization represents direct overhead for them; how big is another question ) . Moreover quantum chemistry application are important subset of jobs a typical HPC cluster load which increase utilization of the particular HPC cluster. this category of jobs also tend to overheat CPUs. The fact that this category of researcher often abuse their privilege to run jobs and schedule too many cores usually is swiped under the carpet. As well as the fact that some of those researchers understand so little in the software then use that their results should be  taken with a grain of salt in any case.

For genomic researcher containers are OK as long underlying hardware does not overheat and throttle CPUs (which actually is not that easy to detect in cloud environment as you do not have direct access to corresponding logs), and if somebody setup them and show how to run applications in this environment. But  their problem are with the virtualized I/O. They need all I/O bandwidth they can get and that's a problem is I/O is virtualized.  Again, this category of researcher they often use some stages that run Python in single threaded node and in attempt to run more jobs simultaneously they often specify minimum number of cores for such jobs. In this case multiple such jobs running on  a single computational node tend to overheat CPUs and throttle speed down considerably.  They only economical way for such researchers to run jobs in the cloud is to use spare capacity and reduced rates for spare capacity on AWS.  This mode is used by several genomic companies which mix "in house computation" that have higher priority and AWS computation which have lower priority and can wait for the "windows of opportunity" to run at reduced rates. the problem here is storage of result of genomic data as they are are bulky as well as the cost of transmitting them back and forth to and from the cloud.  But you can use "cold storage" for non critical results which is much  cheaper. 

NEWS CONTENTS

Old News ;-)

Recommended Links

Softpanorama Recommended

Top articles

Sites

Etc

Society

Groupthink : Two Party System as Polyarchy : Corruption of Regulators : Bureaucracies : Understanding Micromanagers and Control Freaks : Toxic Managers :   Harvard Mafia : Diplomatic Communication : Surviving a Bad Performance Review : Insufficient Retirement Funds as Immanent Problem of Neoliberal Regime : PseudoScience : Who Rules America : Neoliberalism  : The Iron Law of Oligarchy : Libertarian Philosophy

Quotes

War and Peace : Skeptical Finance : John Kenneth Galbraith :Talleyrand : Oscar Wilde : Otto Von Bismarck : Keynes : George Carlin : Skeptics : Propaganda  : SE quotes : Language Design and Programming Quotes : Random IT-related quotes :  Somerset Maugham : Marcus Aurelius : Kurt Vonnegut : Eric Hoffer : Winston Churchill : Napoleon Bonaparte : Ambrose Bierce :  Bernard Shaw : Mark Twain Quotes

Bulletin:

Vol 25, No.12 (December, 2013) Rational Fools vs. Efficient Crooks The efficient markets hypothesis : Political Skeptic Bulletin, 2013 : Unemployment Bulletin, 2010 :  Vol 23, No.10 (October, 2011) An observation about corporate security departments : Slightly Skeptical Euromaydan Chronicles, June 2014 : Greenspan legacy bulletin, 2008 : Vol 25, No.10 (October, 2013) Cryptolocker Trojan (Win32/Crilock.A) : Vol 25, No.08 (August, 2013) Cloud providers as intelligence collection hubs : Financial Humor Bulletin, 2010 : Inequality Bulletin, 2009 : Financial Humor Bulletin, 2008 : Copyleft Problems Bulletin, 2004 : Financial Humor Bulletin, 2011 : Energy Bulletin, 2010 : Malware Protection Bulletin, 2010 : Vol 26, No.1 (January, 2013) Object-Oriented Cult : Political Skeptic Bulletin, 2011 : Vol 23, No.11 (November, 2011) Softpanorama classification of sysadmin horror stories : Vol 25, No.05 (May, 2013) Corporate bullshit as a communication method  : Vol 25, No.06 (June, 2013) A Note on the Relationship of Brooks Law and Conway Law

History:

Fifty glorious years (1950-2000): the triumph of the US computer engineering : Donald Knuth : TAoCP and its Influence of Computer Science : Richard Stallman : Linus Torvalds  : Larry Wall  : John K. Ousterhout : CTSS : Multix OS Unix History : Unix shell history : VI editor : History of pipes concept : Solaris : MS DOS :  Programming Languages History : PL/1 : Simula 67 : C : History of GCC development :  Scripting Languages : Perl history   : OS History : Mail : DNS : SSH : CPU Instruction Sets : SPARC systems 1987-2006 : Norton Commander : Norton Utilities : Norton Ghost : Frontpage history : Malware Defense History : GNU Screen : OSS early history

Classic books:

The Peter Principle : Parkinson Law : 1984 : The Mythical Man-Month :  How to Solve It by George Polya : The Art of Computer Programming : The Elements of Programming Style : The Unix Hater’s Handbook : The Jargon file : The True Believer : Programming Pearls : The Good Soldier Svejk : The Power Elite

Most popular humor pages:

Manifest of the Softpanorama IT Slacker Society : Ten Commandments of the IT Slackers Society : Computer Humor Collection : BSD Logo Story : The Cuckoo's Egg : IT Slang : C++ Humor : ARE YOU A BBS ADDICT? : The Perl Purity Test : Object oriented programmers of all nations : Financial Humor : Financial Humor Bulletin, 2008 : Financial Humor Bulletin, 2010 : The Most Comprehensive Collection of Editor-related Humor : Programming Language Humor : Goldman Sachs related humor : Greenspan humor : C Humor : Scripting Humor : Real Programmers Humor : Web Humor : GPL-related Humor : OFM Humor : Politically Incorrect Humor : IDS Humor : "Linux Sucks" Humor : Russian Musical Humor : Best Russian Programmer Humor : Microsoft plans to buy Catholic Church : Richard Stallman Related Humor : Admin Humor : Perl-related Humor : Linus Torvalds Related humor : PseudoScience Related Humor : Networking Humor : Shell Humor : Financial Humor Bulletin, 2011 : Financial Humor Bulletin, 2012 : Financial Humor Bulletin, 2013 : Java Humor : Software Engineering Humor : Sun Solaris Related Humor : Education Humor : IBM Humor : Assembler-related Humor : VIM Humor : Computer Viruses Humor : Bright tomorrow is rescheduled to a day after tomorrow : Classic Computer Humor

The Last but not Least Technology is dominated by two types of people: those who understand what they do not manage and those who manage what they do not understand ~Archibald Putt. Ph.D

The statements, views and opinions presented on this web page are those of the author (or referenced source) and are not endorsed by, nor do they necessarily reflect, the opinions of the Softpanorama society. We do not warrant the correctness of the information provided or its fitness for any purpose. The site uses AdSense so you need to be aware of Google privacy policy. You you do not want to be tracked by Google please disable Javascript for this site. This site is perfectly usable without Javascript.

Last modified: July 09, 2021