high_scalability high_scalability-2014 knowledge-graph by maker-knowledge-mining
1 high scalability-2014-06-05-Cloud Architecture Revolution
Introduction: The introduction of cloud technologies is not a simple evolution of existing ones, but a real revolution. Like all revolutions, it changes the points of views and redefines all the meanings. Nothing is as before. This post wants to analyze some key words and concepts, usually used in traditional architectures, redefining them according the standpoint of the cloud. Understanding the meaning of new words is crucial to grasp the essence of a pure cloud architecture. << There is no greater impediment to the advancement of knowledge than the ambiguity of words. >> THOMAS REID, Essays on the Intellectual Powers of Man Nowadays, it is required to challenge the limits of traditional architectures that go beyond the normal concepts of scalability and support millions of users (What's Up 500 Million) billions of transactions per day (Salesforce 1.3 billion), five 9s of availability (99.999 AOL). I wish all of you the success of the examples cited above, but do not think that it is co
2 high scalability-2014-05-23-Gone Fishin' 2014
Introduction: Well, not exactly Fishin', but I'll be on a month long vacation starting today. I won't be posting new content, so we'll all have a break. Disappointing, I know. If you've ever wanted to write an article for HighScalability this would be a great time :-) I'd be very interested in your experiences with containers vs VMs if you have some thoughts on the subject. So if the spirit moves you, please write something. See you on down the road...
3 high scalability-2014-05-21-9 Principles of High Performance Programs
Introduction: Arvid Norberg on the libtorrent blog has put together an excellent list of principles of high performance programs , obviously derived from hard won experience programming on bittorrent: Two fundamental causes of performance problems: Memory Latency . A big performance problem on modern computers is the latency of SDRAM. The CPU waits idle for a read from memory to come back. Context Switching . When a CPU switches context "the memory it will access is most likely unrelated to the memory the previous context was accessing. This often results in significant eviction of the previous cache, and requires the switched-to context to load much of its data from RAM, which is slow." Rules to help balance the forces of evil: Batch work . Avoid context switching by batching work. For example, there are vector versions of system calls like writev() and readv() that operate on more than one item per call. An implication is that you want to merge as many writes as possible.
4 high scalability-2014-05-20-It's Networking. In Space! Or How E.T. Will Phone Home.
Introduction: What will the version of the Internet that follows us to the stars look like? Yes, people are really thinking seriously about this sort of thing. Specifically the InterPlanetary Networking Special Interest Group (IPNSIG). Ansible-like faster-than-light communication it isn't. There's no magical warp drive. Nor is a network of telepaths acting as a 'verse spanning telegraph system. It's more mundane than that. And in many ways more interesting as it's sort of like the old Internet on steroids, the one that was based on on UUCP and dial-up connections, but over vastly longer distances and with much longer delays : The Interplanetary Internet (based on IPN, also called InterPlaNet) is a conceived computer network in space, consisting of a set of network nodes which can communicate with each other.[1][2] Communication would be greatly delayed by the great interplanetary distances, so the IPN needs a new set of protocols and technology that are tolerant to large delays and
Introduction: How does the Wayback Machine work? Now with over 400 billion webpages indexed , allowing the Internet to be browsed all the way back to 1996, it's an even more compelling question. I've looked several times but I've never found a really good answer. Here's some information from a thread on Hacker News. It starts with mmagin , a former Archive employee: I can't speak to their current infrastructure (though more of it is open source now - http://archive-access.sourceforge.net/projects/wayback/ ), but as far as the wayback machine, there was no SQL database anywhere in it. For the purposes of making the wayback machine go: Archived data was in ARC file format (predecessor to http://en.wikipedia.org/wiki/Web_ARChive) which is essentially a concatenation of separately gzipped records. That is, you can seek to a particular offset and start decompressing a record. Thus you could get at any archived web page with a triple (server, filename, file-offset) Thus it was spread
6 high scalability-2014-05-16-Stuff The Internet Says On Scalability For May 16th, 2014
Introduction: Hey, it's HighScalability time: Cross Section of an Undersea Cable. It's industrial art. The parts . The story . 400,000,000,000 : Wayback Machine pages indexed; 100 billion : Google searches per month; 10 million : Snapchat monthly user growth. Quotable Quotes: @duncanjw : The Great Rewrite - many apps will be rewritten not just replatformed over next 10 years says @cote #openstacksummit @RFFlores : The Openstack conundrum. If you don't adopt it, you will regret it in the future. If you do adopt it, you will regret it now elementai : I love Redis so much, it became like a superglue where "just enough" performance is needed to resolve a bottleneck problem, but you don't have resources to rewrite a whole thing in something fast. @antirez : "when software engineering is reduced to plumbing together generic systems, software engineers lose their sense of ownership" Tom Akehurst : Microservices vs. monolith is a false di
Introduction: So how do you knit multiple datacenters and many thousands of phones and other clients into a single cooperating system? Usually you don't. It's too hard. We see nascent attempts in services like Firebase and Parse. SwiftCloud , as described in SwiftCloud: Fault-Tolerant Geo-Replication Integrated all the Way to the Client Machine , goes two steps further, by leveraging Conflict free Replicated Data Types (CRDTs), which means "data can be replicated at multiple sites and be updated independently with the guarantee that all replicas converge to the same value. In a cloud environment, this allows a user to access the data center closer to the user, thus optimizing the latency for all users." While we don't see these kind of systems just yet, they are a strong candidate for how things will work in the future, efficiently using resources at every level while supporting huge numbers of cooperating users. Abstract : Client-side logic and storage are increasingly used in web a
Introduction: After Google cut prices on their Google Cloud Platform Amazon quickly followed with their own price cuts . Even more interesting is what the future holds for pricing. The near future looks great. After that? We'll see. Adrian Cockcroft highlights that Google thinks prices should follow Moore’s law, which means we should expect prices to halve every 18-24 months. That's good news. Greater cost certainty means you can make much more aggressive build out plans. With the savings you can hire more people, handle more customers, and add those media rich features you thought you couldn't afford. Design is directly related to costs. Without Google competing with Amazon there's little doubt the price reduction curve would be much less favorable. As a late cloud entrant Google is now in a customer acquisition phase, so they are willing to pay for customers, which means lower prices are an acceptable cost of doing business. Profit and high margins are not the objective. Getting marke
Introduction: This is a guest repost by Venkatesh CM at Architecture Issues Scaling Web Applications . I will cover architecture issues that show up while scaling and performance tuning large scale web application in this blog. Lets start by defining few terms to create common understanding and vocabulary. Later on I will go through different issues that pop-up while scaling web application like Architecture bottlenecks Scaling Database CPU Bound Application IO Bound Application Determining optimal thread pool size of an web application will be covered in next blog. Performance Term performance of web application is used to mean several things. Most developers are primarily concerned with are response time and scalability. Response Time Is the time taken by web application to process request and return response. Applications should respond to requests (response time) within acceptable duration. If application is taking beyond the acceptable time, it is said to
10 high scalability-2014-05-09-Stuff The Internet Says On Scalability For May 9th, 2014
Introduction: Hey, it's HighScalability time: NASA captures Guatemala volcano erupting from space 40,000 exabytes : from now until 2020, the digital universe will about double every two years; $650,000 : amount raised by the MaydayPAC in one week. Quotable Quotes: @BenedictEvans : Masayoshi Son: $20m initial investment in Alibaba, current stake worth $58bn. @iamdevloper : I sneezed earlier and Siri compiled it to valid Perl. @cdixon : "There is not enough competition in the last mile market to allow a true market to function" @PatrickMcFadin : Get ready for some serious server density. AMD is working on K12, brand-new x86 and ARM cores. This plus 8T SSD? With age comes changing priorities. Facebook is now 10 and has grown up . They are no longer moving fast and breaking things. They are now into the stability thing. Letting developers know they are a stable platform. The play is to get all that beautiful data from developers by bei
11 high scalability-2014-05-07-Update on Disqus: It's Still About Realtime, But Go Demolishes Python
Introduction: Our last article on Disqus: How Disqus Went Realtime With 165K Messages Per Second And Less Than .2 Seconds Latency , was a little out of date, but the folks at Disqus have been busy implementing, not talking, so we don't know a lot about what they are doing now, but we do have a short update in C1MM and NGINX by John Watson and an article Trying out this Go thing . So Disqus has grown a bit: 1.3 billion unique visitors 10 billion page views 500 million users engaged in discussions 3 million communities 25 million comments They are still all about realtime, but Go replaced Python in their Realtime system: Original Realtime backend was written in a pretty lightweight Python + gevent. The realtime service is a hybrid of CPU intensive tasks + lots of network IO. Gevent was handling the network IO without an issue, but at higher contention, the CPU was choking everything. Switching over to Go removed that contention, which was the primary issue that was being se
Introduction: This a guest post by Rajkumar Iyer , a Member of Technical Staff at Aerospike. About a year ago, Aerospike embarked upon a quest to increase in-memory database performance - 1 Million TPS on a single inexpensive commodity server. NoSQL has the reputation of speed, and we saw great benefit from improving latency and throughput of cacheless architectures. At that time, we took a version of Aerospike delivering about 200K TPS, improved a few things - performance went to 500k TPS - and published the Aerospike 2.0 Community Edition. We then used kernel tuning techniques and published the recipe for how we achieved 1 M TPS on $5k of hardware. This year we continued the quest. Our goal was to achieve 1 Million database transactions per second per server; more than doubling previous performance. This compares to Cassandra’s boast of 1M TPS on over 300 servers in Google Compute Engine - at a cost of $2 million dollars per year. We achieved this without kernel tuning. This article d
13 high scalability-2014-05-02-Stuff The Internet Says On Scalability For May 2nd, 2014
Introduction: Hey, it's HighScalability time: Google's new POWER8 server motherboard 1 trillion : number of scents your nose can smell; millions of square feet : sprawling new server farms Quotable Quotes: Gideon Lewis-Kraus : As the engineer and writer Alex Payne put it, these startups represent “the field offices of a large distributed workforce assembled by venture capitalists and their associate institutions,� doing low-overhead, low-risk R&D; for five corporate giants. In such a system, the real disillusionment isn’t the discovery that you’re unlikely to become a billionaire; it’s the realization that your feeling of autonomy is a fantasy, and that the vast majority of you have been set up to fail by design. @aphyr : I was already sold on immutability, pure functions, combinators, etc. What forced me out of Haskell was the impene
14 high scalability-2014-05-01-Paper: Can Programming Be Liberated From The Von Neumann Style?
Introduction: Famous computer scientist John Backus , he's the B in BNF(Backus-Naur form) and the creator of Fortran, gave a Turing Award Lecture titled Can programming be liberated from the von Neumann style?: a functional style and its algebra of programs , that has layed out a division in programming that lives long after it was published in 1977. It's the now familiar argument for why functional programming is superior: The assignment statement is the von Neumann bottleneck of programming languages and keeps us thinking in word-at-a-time terms in much the same way the computer's bottleneck does. ... The second world of conventional programming languages is the world of statements. The primary statement in that world is the assignment statement itself. All the other statements of the language exist in order to make it possible to perform a computation that must be based on this primitive construct: the assignment statement. Here's a response by Dijkstra A review of the 1977 Turi
15 high scalability-2014-04-30-10 Tips for Optimizing NGINX and PHP-fpm for High Traffic Sites
Introduction: Adrian Singer has boiled down 7 years of experience to a set of 10 very useful tips on how to best optimize NGINX and PHP-fpm for high traffic sites : Switch from TCP to UNIX domain sockets . When communicating to processes on the same machine UNIX sockets have better performance the TCP because there's less copying and fewer context switches. Adjust Worker Processes . Set the worker_processes in your nginx.conf file to the number of cores your machine has and increase the number of worker_connections. Setup upstream load balancing . Multiple upstream backends on the same machine produce higher throughout than a single one. Disable access log files . Log files on high traffic sites involve a lot of I/O that has to be synchronized across all threads. Can have a big impact. Enable GZip . Cache information about frequently accessed files . Adjust client timeouts . Adjust output buffers . /etc/sysctl.conf tuning . Monitor . Continually monitor the number
Introduction: Who's Hiring? Apple is hiring a Senior Engineer in their Mobile Services team. We seek an accomplished server-side engineer capable of delivering an extraordinary portfolio of features and services based on emerging technologies to our internal customers. Please apply here . Apple is hiring a Software Engineer in their Messaging Services team. We build the cloud systems that power some of the busiest applications in the world, including iMessage, FaceTime and Apple Push Notifications. You'll have the opportunity to explore a wide range of technologies, developing the server software that is driving the future of messaging and mobile services. Please apply here . Apple is hiring an Enterprise Software Engineer. Apple's Emerging Technology Services group provides a Java based SOA platform for various applications to interact with each other. The platform is designed to handle millions of messages a day with very low latency. We have an immediate opening for a
Introduction: Here's an Update On Disqus: It's Still About Realtime, But Go Demolishes Python . How do you add realtime functionality to a web scale application? That's what Adam Hitchcock , a Software Engineer at Disqus talks about in an excellent talk: Making DISQUS Realtime ( slides ). Disqus had to take their commenting system and add realtime capabilities to it. Not something that's easy to do when at the time of the talk (2013) they had had just hit a billion unique visitors a month. What Disqus developed is a realtime commenting system called “realertime” that was tested to handle 1.5 million concurrently connected users, 45,000 new connections per second, 165,000 messages/second, with less than .2 seconds latency end-to-end. The nature of a commenting system is that it is IO bound and has a high fanout, that is a comment comes in and must be sent out to a lot of readers. It's a problem very similar to what Twitter must solve . Disqus' solution was quite interesting as was th
18 high scalability-2014-04-25-Stuff The Internet Says On Scalability For April 25th, 2014
Introduction: Hey, it's HighScalability time: New World Record BASE jumping from World's Tallest Building. #crazy 30 billion : total Pinterest pins; 500,000,000 : What'sApp users (700 million photos and 100 million videos every single day); 1 billion : Facebook active users on phones and tablets. Quotable Quotes: @jimplush : Google spent 2.3 billion on infrastructure in Q1. Remember that when you say you want to be "the Google of something" Clay Shirky: I think one of the things that happened to the P2P market for infrastructure is that users preference for predictable pricing vs resource-sensitive pricing is so overwhelming that they will overpay to anyone who can promise flat prices. And because the logic of centralization vs decentralization is so price sensitive, I don't think there is any logical reason to assume a broadly stable class of apps, separate from c
Introduction: How is it possible that a wooden Shinto shrine built in the 7th century is still standing? The answer depends on how you answer this philosophical head scratcher: With nearly every cell in your body continually being replaced, are you still the same person? The Ise Grand Shrine has been in continuous existence for over 1300 years because every twenty years an exact replica has been rebuilt on an adjacent footprint. The former temple is then dismantled. Now that's resilience. If you want something to last make it a living part of a culture. It's not so much the building that is remade, what is rebuilt and passed down from generation to generation is the meme that the shrine is important and worth preserving. The rest is an unfolding of that imperative. You can see echoes of this same process in Open Source projects like Linux and the libraries and frameworks that get themselves reconstructed in each new environment. The patterns of recurrence in software are the result of Darw
20 high scalability-2014-04-21-This is why Microsoft won. And why they lost.
Introduction: My favorite kind of histories are those told from an insider's perspective. The story of Richard the Lionheart is full of great battles and dynastic intrigue. The story of one of his soldiers, not so much. Yet the soldiers' story, as someone who has experienced the real consequences of decisions made and actions taken, is more revealing. We get such a history in Chat Wars , a wonderful article written by David Auerbach, who in 1998 worked at Microsoft on MSN Messenger Service, Microsoft’s instant messaging app (for a related story see The Rise and Fall of AIM, the Breakthrough AOL Never Wanted ). It's as if Herodotus visited Microsoft and wrote down his experiences. It has that same sort of conversational tone, insightful on-the-ground observations, and facts no outsider might ever believe. Much of the article is a play-by-play account of the cat and mouse game David plays changing Messenger to track AOL's Instant Messenger protocol changes. AOL repeatedly tried to make it so M
21 high scalability-2014-04-18-Stuff The Internet Says On Scalability For April 18th, 2014
22 high scalability-2014-04-16-Six Lessons Learned the Hard Way About Scaling a Million User System
24 high scalability-2014-04-14-How do you even do anything without using EBS?
25 high scalability-2014-04-11-Stuff The Internet Says On Scalability For April 11th, 2014
27 high scalability-2014-04-08-Microservices - Not a free lunch!
29 high scalability-2014-04-04-Stuff The Internet Says On Scalability For April 4th, 2014
30 high scalability-2014-04-03-Leslie Lamport to Programmers: You're Doing it Wrong
31 high scalability-2014-04-01-The Mullet Cloud Selection Pattern
34 high scalability-2014-03-28-Stuff The Internet Says On Scalability For March 28th, 2014
35 high scalability-2014-03-27-Strategy: Cache Stored Procedure Results
38 high scalability-2014-03-21-Stuff The Internet Says On Scalability For March 21st, 2014
43 high scalability-2014-03-14-Stuff The Internet Says On Scalability For March 14th, 2014
44 high scalability-2014-03-12-Paper: Scalable Eventually Consistent Counters over Unreliable Networks
45 high scalability-2014-03-11-Douglas Adams - 3 Rules that Describe Our Reactions to Technologies
47 high scalability-2014-03-10-Let's Play a Game of Take It or Leave It - Game 1
48 high scalability-2014-03-07-Stuff The Internet Says On Scalability For March 7th, 2014
49 high scalability-2014-03-05-10 Things You Should Know About Running MongoDB at Scale
52 high scalability-2014-02-28-Stuff The Internet Says On Scalability For February 28th, 2014
53 high scalability-2014-02-26-The WhatsApp Architecture Facebook Bought For $19 Billion
54 high scalability-2014-02-25-Peter Norvig's 9 Master Steps to Improving a Program
55 high scalability-2014-02-21-Stuff The Internet Says On Scalability For February 21st, 2014
59 high scalability-2014-02-14-Stuff The Internet Says On Scalability For February 14th, 2014
60 high scalability-2014-02-13-Snabb Switch - Skip the OS and Get 40 million Requests Per Second in Lua
61 high scalability-2014-02-12-Paper: Network Stack Specialization for Performance
63 high scalability-2014-02-07-Stuff The Internet Says On Scalability For February 7th, 2014
66 high scalability-2014-02-03-How Google Backs Up the Internet Along With Exabytes of Other Data
67 high scalability-2014-01-31-Stuff The Internet Says On Scalability For January 31st, 2014
68 high scalability-2014-01-29-10 Things Bitly Should Have Monitored
70 high scalability-2014-01-24-Stuff The Internet Says On Scalability For January 24th, 2014
73 high scalability-2014-01-20-8 Ways Stardog Made its Database Insanely Scalable
74 high scalability-2014-01-17-Stuff The Internet Says On Scalability For January 17th, 2014
76 high scalability-2014-01-14-SharePoint VPS solution
77 high scalability-2014-01-14-Ask HS: Design and Implementation of scalable services?
78 high scalability-2014-01-13-NYTimes Architecture: No Head, No Master, No Single Point of Failure
79 high scalability-2014-01-10-Stuff The Internet Says On Scalability For January 10th, 2014
80 high scalability-2014-01-08-Under Snowden's Light Software Architecture Choices Become Murky
83 high scalability-2014-01-03-Stuff The Internet Says On Scalability For January 3rd, 2014
84 high scalability-2014-01-02-xkcd: How Standards Proliferate: