load shedding

8 results back to index

pages: 719 words: 181,090

Site Reliability Engineering by Betsy Beyer, Chris Jones, Jennifer Petoff, Niall Richard Murphy

Air France Flight 447, anti-pattern, barriers to entry, business intelligence, business process, Checklist Manifesto, cloud computing, combinatorial explosion, continuous integration, correlation does not imply causation, crowdsourcing, database schema, defense in depth, DevOps, en.wikipedia.org, fault tolerance, Flash crash, George Santayana, Google Chrome, Google Earth, job automation, job satisfaction, linear programming, load shedding, loose coupling, meta analysis, meta-analysis, minimum viable product, MVC pattern, performance metric, platform as a service, revision control, risk tolerance, side project, six sigma, the scientific method, Toyota Production System, trickle-down economics, web application, zero day

Such strategies are more likely to be needed for shared services. Graceful degradation takes the concept of load shedding one step further by reducing the amount of work that needs to be performed. In some applications, it’s possible to significantly decrease the amount of work or time needed by decreasing the quality of responses. For instance, a search application might only search a subset of data stored in an in-memory cache rather than the full on-disk database or use a less-accurate (but faster) ranking algorithm when overloaded. When evaluating load shedding or graceful degradation options for your service, consider the following: Which metrics should you use to determine when load shedding or graceful degradation should kick in (e.g,. CPU usage, latency, queue length, number of threads used, whether your service enters degraded mode automatically or if manual intervention is necessary)?

time requirements, Balance in Quality training for, Learning Paths That Are Cumulative and Orderly-A Hunger for Failure: Reading and Sharing Postmortems training materials, Creating Stellar Reverse Engineers and Improvisational Thinkers typical activities, Life of an On-Call Engineer one-phase pipelines, Initial Effect of Big Data on the Simple Pipeline Pattern open commenting/annotation system, Collaborate and Share Knowledge operational loadcross-industry lessons, Automating Away Repetitive Work and Operational Overhead managing, Managing Operational Load ongoing responsibilities, Managing Operational Load types of, Dealing with Interrupts operational overload, Operational Overload operational underload, A Treacherous Enemy: Operational Underload operational work (see toil) out-of-band checks and balances, Choosing a Strategy for Superior Data Integrity, Out-of-band data validation out-of-band communications systems, What went well outage trackingbaselines and progress tracking, Tracking Outages benefits of, Unexpected Benefits Escalator, Escalator Outalator, Outalator-Reporting and communication Outalatoraggregation in, Aggregation benefits of, Outalator building your own, Outalator incident analysis, Analysis notification process, Outalator reporting and communication, Reporting and communication tagging in, Tagging overhead, Toil Defined overload handlingapproaches to, Handling Overload best practices for, Overloads and Failure client-side throttling, Client-Side Throttling load from connections, Load from Connections overload errors, Handling Overload Errors overview of, Conclusions per-client retry budget, Deciding to Retry per-customer limits, Per-Customer Limits per-request retry budget, Deciding to Retry product launches and, Overload Behavior and Load Tests request criticality, Criticality retrying requests, Deciding to Retry(see also retries, RPC) utilization signals, Utilization Signals(see also cascading failures) P package managers, Packaging packet encapsulation, Load Balancing at the Virtual IP Address Paxos consensus algorithmClassic Paxos algorithm, Reasoning About Performance: Fast Paxos disk access and, Disk Access Egalitarian Paxos consensus algorithm, Stable Leaders Fast Paxos consensus algorithm, Reasoning About Performance: Fast Paxos, The Use of Paxos Lamport’s Paxos protocol, How Distributed Consensus Works(see also consensus algorithms) performanceefficiency and, Efficiency and Performance monitoring, Worrying About Your Tail (or, Instrumentation and Performance) performance tests, System tests periodic pipelines, Challenges with the Periodic Pipeline Pattern periodic scheduling (see cron) persistent storage, Disk Access Photon, Number of Replicas pipelining, Batching planned changes, Planned Changes, Drains, or Turndowns policies and procedures, enforcing, Enforcement of Policies and Procedures post hoc analysis, Setting Reasonable Expectations for Monitoring postmortemsbenefits of, Postmortem Culture: Learning from Failure best practices for, Google’s Postmortem Philosophy-Introducing a Postmortem Culture, Postmortems collaboration and sharing in, Collaborate and Share Knowledge concept of, Postmortem Culture: Learning from Failure cross-industry lessons, Postmortem Culture-Postmortem Culture example postmortem, Example Postmortem-Timeline formal review and publication of, Collaborate and Share Knowledge Google's philosophy for, Google’s Postmortem Philosophy guidelines for, Ensuring a Durable Focus on Engineering introducing postmortem cultures, Introducing a Postmortem Culture on-call engineering and, A Hunger for Failure: Reading and Sharing Postmortems ongoing improvements to, Conclusion and Ongoing Improvements rewarding participation in, Introducing a Postmortem Culture triggers for, Google’s Postmortem Philosophy privacy, Choosing a Strategy for Superior Data Integrity proactive testing, Encourage Proactive Testing problem reports, Problem Report process death, Process Death process health checks, Stop Health Check Failures/Deaths process updates, Process Updates process-induced emergencies, Process-Induced Emergency Prodtest (Production Test), Detecting Inconsistencies with Prodtest product launchesbest practices for, Progressive Rollouts defined, Reliable Product Launches at Scale development of Launch Coordination Engineering (LCE), Development of LCE-Infrastructure churn driving convergence and simplification, Driving Convergence and Simplification launch coordination checklists, The Launch Checklist-Example action items, Launch Coordination Checklist launch coordination engineering, Launch Coordination Engineering NORAD Tracks Santa example, Reliable Product Launches at Scale overview of, Conclusion processes for, Setting Up a Launch Process rate of, Reliable Product Launches at Scale techniques for reliable, Selected Techniques for Reliable Launches-Overload Behavior and Load Tests production environment (see Google production environment) production inconsistenciesdetecting with Prodtest, Detecting Inconsistencies with Prodtest resolving idempotently, Resolving Inconsistencies Idempotently production meetings, Communications: Production Meetings-Attendanceagenda example, Example Production Meeting Minutes production probes, Production Probes Production Readiness Review process (see SRE engagement model) production tests, Production Tests protocol buffers (protobufs), Our Software Infrastructure, Integration Protocol Data Units, Load Balancing at the Virtual IP Address provisioning, guidelines for, Provisioning PRR (Production Readiness Review) model, The PRR Model, Production Readiness Reviews: Simple PRR Model-Continuous Improvement push frequency, Motivation for Error Budgets push managers, Ongoing responsibilities Python’s safe_load, Integration Q “queries per second” model, The Pitfalls of “Queries per Second” Query of Death, Process Death queuingcontrolled delay, Load Shedding and Graceful Degradation first-in, first-out, Load Shedding and Graceful Degradation last-in, first-out, Load Shedding and Graceful Degradation management of, Queue Management, Reliable Distributed Queuing and Messaging queuing-as-work-distribution pattern, Reliable Distributed Queuing and Messaging quorum (see distributed consensus systems) R Raft consensus protocol, Multi-Paxos: Detailed Message Flow, Stable Leaders(see also consensus algorithms) RAID, Overarching Layer: Replication Rapid automated release system, Continuous Build and Deployment, Rapid read workload, scaling, Scaling Read-Heavy Workloads real backups, Backups Versus Archives real-time collaboration, Collaborate and Share Knowledge recoverability, Challenges of Maintaining Data Integrity Deep and Wide recovery, Knowing That Data Recovery Will Work recovery systems, Delivering a Recovery System, Rather Than a Backup System recursion (see recursion) recursive DNS servers, Load Balancing Using DNS recursive separation of responsibilities, Recursive Separation of Responsibilities redundancy, Challenges of Maintaining Data Integrity Deep and Wide, Overarching Layer: Replication Reed-Solomon erasure codes, Overarching Layer: Replication regression tests, System tests release engineeringchallenges of, Release Engineering continuous build and deployment, Continuous Build and Deployment-Configuration Management defined, Release Engineering instituting, Start Release Engineering at the Beginning philosophy of, Philosophy-Enforcement of Policies and Procedures the role of release engineers, The Role of a Release Engineer wider application of, Conclusions reliability testingamount required, Testing for Reliability benefits of, Conclusion break-glass mechanisms, Expect Testing Fail canary tests, Canary test configuration tests, Configuration test coordination of, The Need for Speed creating test and build environments, Creating a Test and Build Environment error budgets, Pursuing Maximum Change Velocity Without Violating a Service’s SLO, Motivation for Error Budgets-Forming Your Error Budget, Error Budgets expecting test failure, Expect Testing Fail-Expect Testing Fail fake backend versions, Production Probes goals of, Testing for Reliability importance of, Preface integration tests, Integration tests, Integration MTTR and, Testing for Reliability performance tests, System tests proactive, Encourage Proactive Testing production probes, Production Probes production tests, Production Tests regression tests, System tests reliability goals, Embracing Risk sanity testing, System tests segregated environments and, Pushing to Production smoke tests, System tests speed of, The Need for Speed statistical tests, Testing Disaster stress tests, Stress test system tests, System tests testing at scale, Testing at Scale-Production Probes timing of, Production Tests unit tests, Unit tests reliable replicated datastores, Reliable Replicated Datastores and Configuration Stores Remote Procedure Call (RPC), Our Software Infrastructure, Examine, Criticalitybimodal, Bimodal latency deadlinesmissing, Missing deadlines propagating, Load Shedding and Graceful Degradation, Deadline propagation queue management, Queue Management, Reliable Distributed Queuing and Messaging selecting, Latency and Deadlines retries, Retries-Retries RPC criticality, Criticality(see also overload handling) replicasadding, Capacity and Load Balancing drawbacks of leader replicas, Capacity and Load Balancing location of, Location of Replicas, Quorum composition number deployed, Number of Replicas replicated logs, Number of Replicas replicated state machine (RSM), Reliable Replicated State Machines replication, Challenges of Maintaining Data Integrity Deep and Wide, Overarching Layer: Replication request latency, Indicators, The Four Golden Signals request profile changes, Request profile changes request success rate, Measuring Service Risk resilience testing, Practices resourcesallocation of, Hardware, Managing Machines exhaustion, Resource Exhaustion limits, Resource limits(see also capacity planning) restores, 1T Versus 1E: Not “Just” a Bigger Backup retention, Retention retries, RPCavoiding, Deciding to Retry cascading failures due to, Retries considerations for automatic, Retries diagnosing outages due to, Retries handling overload errors and, Handling Overload Errors per-client retry budgets, Deciding to Retry per-request retry budgets, Deciding to Retry reverse engineering, Reverse Engineers: Figuring Out How Things Work reverse proxies, What went well revision history, First Layer: Soft Deletion risk managementbalancing risk and innovation, Embracing Risk costs of, Managing Risk error budgets, Motivation for Error Budgets-Benefits, Error Budgets key insights, Benefits measuring service risk, Measuring Service Risk risk tolerance of services, Risk Tolerance of Services-Example: Frontend infrastructure rollback procedures, What we learned rollouts, New Rollouts, Rollout Planning, Progressive Rollouts root causeanalysis of, Practices, Google’s Postmortem Philosophy(see also postmortems) defined, Definitions Round Robin policy, Simple Round Robin round-trip-time (RTT), Distributed Consensus Performance and Network Latency rows, Hardware rule evaluation, in monitoring systems, Rule Evaluation-Rule Evaluation S Safari® Books Online, Safari® Books Online sanity testing, System tests saturation, The Four Golden Signals scaledefined, Choosing a Strategy for Superior Data Integrity issues in, Scaling issues: Fulls, incrementals, and the competing forces of backups and restores securityin release engineering, Enforcement of Policies and Procedures new approach to, Practices self-service model, Self-Service Model separation of responsibilities, Recursive Separation of Responsibilities serversvs. clients, Our Software Infrastructure defined, Hardware overload scenario, Server Overload preventing overload, Preventing Server Overload-Always Go Downward in the Stack service availabilityavailability table, Availability Table cost factors, Cost, Cost defined, Indicators target for consumer services, Target level of availability target for infrastructure service, Target level of availability time-based equation, Measuring Service Risk types of consumer service failures, Types of failures types of infrastructure services failures, Types of failures service health checks, Stop Health Check Failures/Deaths service latencylooser approach to, Other service metrics monitoring for, The Four Golden Signals service level agreements (SLAs), Agreements service level indicators (SLIs)aggregating raw measurements, Aggregation collecting indicators, Collecting Indicators defined, Indicators standardizing indicators, Standardize Indicators service level objectives (SLOs)agreements in practice, Agreements in Practice best practices for, Define SLOs Like a User choosing, Service Level Objectives-Objectives control measures, Control Measures defined, Objectives defining objectives, Objectives in Practice selecting relevant indicators, What Do You and Your Users Care About?

Unless you test in a realistic environment, it’s very hard to predict exactly which resource will be exhausted and how that resource exhaustion will manifest. For details, see “Testing for Cascading Failures”. Serve degraded results Serve lower-quality, cheaper-to-compute results to the user. Your strategy here will be service-specific. See “Load Shedding and Graceful Degradation”. Instrument the server to reject requests when overloaded Servers should protect themselves from becoming overloaded and crashing. When overloaded at either the frontend or backend layers, fail early and cheaply. For details, see “Load Shedding and Graceful Degradation”. Instrument higher-level systems to reject requests, rather than overloading servers Note that because rate limiting often doesn’t take overall service health into account, it may not be able to stop a failure that has already begun.

pages: 757 words: 193,541

The Practice of Cloud System Administration: DevOps and SRE Practices for Web Services, Volume 2 by Thomas A. Limoncelli, Strata R. Chalup, Christina J. Hogan

active measures, Amazon Web Services, anti-pattern, barriers to entry, business process, cloud computing, commoditize, continuous integration, correlation coefficient, database schema, Debian, defense in depth, delayed gratification, DevOps, domain-specific language, en.wikipedia.org, fault tolerance, finite state, Firefox, Google Glasses, information asymmetry, Infrastructure as a Service, intermodal, Internet of things, job automation, job satisfaction, load shedding, loose coupling, Malcom McLean invented shipping containers, Marc Andreessen, place-making, platform as a service, premature optimization, recommendation engine, revision control, risk tolerance, side project, Silicon Valley, software as a service, sorting algorithm, statistical model, Steven Levy, supply-chain management, Toyota Production System, web application, Yogi Berra

Shared resource pools are not just appropriate for machines, but may also be used for storage and other resources. Load Shedding Another strategy is load shedding. With this strategy the service turns away some users so that other users can have a good experience. To make an analogy, an overloaded phone system doesn’t suddenly disconnect all existing calls. Instead, it responds to any new attempts to make a call with a “fast busy” tone so that the person will try to make the call later. An overloaded web site should likewise give some users an immediate response, such as a simple “come back later” web page, rather than requiring them to time out after minutes of waiting. A variation of load shedding is stopping certain tasks that can be put off until later. For example, low-priority database updates could be queued up for processing later; a social network that stores reputation points for users might store the fact that points have been awarded rather than processing them; nightly bulk file transfers might be delayed if the network is overloaded.

For example, low-priority database updates could be queued up for processing later; a social network that stores reputation points for users might store the fact that points have been awarded rather than processing them; nightly bulk file transfers might be delayed if the network is overloaded. That said, tasks that can be put off for a couple of hours might cause problems if they are put off forever. There is, after all, a reason they exist. For any activity that is delayed due to load shedding, there must be a plan on how such a delay is handled. Establish a service level agreement (SLA) to determine how long something can be delayed and to identify a timeline of actions that should be undertaken to mitigate problems or extend the deadlines. Low-priority updates might become a high priority after a certain amount of time. If many systems are turned off due to load shedding, it might be possible to enable them, one at a time, to let each catch up. To be able to manage such situations one must have visibility into the system so that prioritization decisions can be made. For example, knowing the age of a task (how long it has been delayed), predicting how long it will take to process, and indicating how close it is to a deadline will permit operations personnel to gauge when delayed items should be continued

DiRT tests, 320 meta-work, 162 packages, 204 test event planning, 319 time management, 256 Linear scaling, 476–477 Link-shortening site example, 87–89 Linking tickets to subsystems, 263–264 Linux in dot-bomb era, 460–461 Live code changes, 236 Live restores, 36 Live schema changes, 234–236 Live service upgrades. See Upgrading live services Load balancers failures, 134–136 first web era, 456 with multiple backend replicas, 12–13 with shared state, 75 three-tier web service, 72–74 Load sharing vs. hot spares, 126 Load shedding, 139 Load testing, 215 Local business listings in Google Maps, 42 Local labor laws, 43 Logarithmic scaling, 476 Logistics in disaster preparedness, 318–320 Logistics team in Incident Command System, 326 Loglinear scaling, 476–477 Logs, 11, 340 approach, 341 design documents, 282 timestamps, 341–342 Long-term analysis, 354 Long-term fixes oncall, 299–300 vs. quick, 295–296 Longitudinal hardware failure study, 133–134 Look-for’s, 407 Lookup-oriented splits in AKF Scaling Cube, 102–104 Loosely coupled systems, 24–25 Lower-latency services in cloud computing era, 469 LREs (Launch Readiness Engineers), 157–158 LRRs (Launch Readiness Reviews), 159 LRU (Least Recently Used) algorithm, 107 Lynch, N., 23 MACD (moving average convergence/divergence) metric, 367, 378–379 MACD signal line, 367 Machine Learning service, 55 Machines automated configuration example, 251 defined, 10 failures, 134 KPI example, 393–396 Madrigal, A.

pages: 464 words: 127,283

Smart Cities: Big Data, Civic Hackers, and the Quest for a New Utopia by Anthony M. Townsend

Amazon: amazon.comamazon.co.ukamazon.deamazon.fr

1960s counterculture, 4chan, A Pattern Language, Airbnb, Amazon Web Services, anti-communist, Apple II, Bay Area Rapid Transit, Burning Man, business process, call centre, carbon footprint, charter city, chief data officer, clean water, cleantech, cloud computing, computer age, congestion charging, connected car, crack epidemic, crowdsourcing, DARPA: Urban Challenge, data acquisition, Deng Xiaoping, digital map, Donald Davies, East Village, Edward Glaeser, game design, garden city movement, Geoffrey West, Santa Fe Institute, George Gilder, ghettoisation, global supply chain, Grace Hopper, Haight Ashbury, Hedy Lamarr / George Antheil, hive mind, Howard Rheingold, interchangeable parts, Internet Archive, Internet of things, Jacquard loom, Jacquard loom, Jane Jacobs, jitney, John Snow's cholera map, Khan Academy, Kibera, knowledge worker, load shedding, M-Pesa, Mark Zuckerberg, megacity, mobile money, mutually assured destruction, new economy, New Urbanism, Norbert Wiener, Occupy movement, off grid, openstreetmap, packet switching, Parag Khanna, patent troll, Pearl River Delta, place-making, planetary scale, popular electronics, RFC: Request For Comment, RFID, ride hailing / ride sharing, Robert Gordon, self-driving car, sharing economy, Silicon Valley, Skype, smart cities, Smart Cities: Big Data, Civic Hackers, and the Quest for a New Utopia, smart grid, smart meter, social graph, social software, social web, special economic zone, Steve Jobs, Steve Wozniak, Stuxnet, supply-chain management, technoutopianism, Ted Kaczynski, telepresence, The Death and Life of Great American Cities, too big to fail, trade route, Tyler Cowen: Great Stagnation, Upton Sinclair, uranium enrichment, urban decay, urban planning, urban renewal, Vannevar Bush, working poor, working-age population, X Prize, Y2K, zero day, Zipcar

Dynamic pricing can dramatically reduce swings in demand for power and increase overall generating efficiency, but load shifting can also be automated and proactive. Smart meters that communicate directly with smart appliances might automatically reschedule a load of wash for later in the day when demand and prices are likely to fall. Even the most sophisticated load-shifting scheme will one day meet its limit. That’s when utilities wield their trump card—load shedding—a kind of targeted blackout. Traditionally, load shedding was a manual process. Utilities would cut deals with large users of electricity like factories and universities to shut down power during peaking crises in return for a discount on their regular rates. Smart meters will allow these miniblackouts to be replaced by sophisticated surgical drawdowns on sacrificial facilities and equipment. A university might agree to have its dormitories or office lighting shut off while service to sensitive laboratory instruments, for instance, is maintained.

While peaking plants can also be highly efficient—most are natural-gas–powered turbines—they are far more costly per unit of power to build and run. If only the peaks could be evened out, fewer peaking plants would be needed and utilities could focus more on ruthlessly fine-tuning base load plants to be as lean and clean as possible.48 Smart grids offer two tricks to even out the peaks: load shifting and load shedding. Load shifting, the gentler of the two, tries to spread demand for power away from peak periods of demand through price incentives. In their simplest form, smart meters allow businesses and consumers to see the true cost of generating electricity during periods of high demand. As they fire up those costly peaking plants, utilities simply pass the higher generating cost along to consumers. Dynamic pricing can dramatically reduce swings in demand for power and increase overall generating efficiency, but load shifting can also be automated and proactive.

Even as demand surges, building new power plants only gets harder as NIMBY-led resistance to plant construction spreads in many countries. The wiggle room that once existed in the form of reserve generating capacity is fast disappearing, raising the possibility of regular blackouts in the future. During the 1990s, demand for electricity grew by 35 percent in the United States, but generating capacity increased by only 18 percent.49 According to Siemens, smart grids will help utility engineers sleep at night, since load shedding and load shifting could reduce national electricity needs by up to 10 percent. 50 Environmentalists will cheer because improved demand management removes a key obstacle to greater reliance on renewable generating sources, which are notoriously unreliable base capacity—the sun doesn’t always shine and the wind doesn’t always blow. Even hydropower generated at dams depends on reliable seasonal rains to fill up rivers.

pages: 540 words: 103,101

Building Microservices by Sam Newman

Amazon: amazon.comamazon.co.ukamazon.deamazon.fr

airport security, Amazon Web Services, anti-pattern, business process, call centre, continuous integration, create, read, update, delete, defense in depth, don't repeat yourself, Edward Snowden, fault tolerance, index card, information retrieval, Infrastructure as a Service, inventory management, job automation, load shedding, loose coupling, platform as a service, premature optimization, pull request, recommendation engine, social graph, software as a service, source of truth, the built environment, web application, WebSocket, x509 certificate

In many ways, bulkheads are the most important of these three patterns. Timeouts and circuit breakers help you free up resources when they are becoming constrained, but bulkheads can ensure they don’t become constrained in the first place. Hystrix allows you, for example, to implement bulkheads that actually reject requests in certain conditions to ensure that resources don’t become even more saturated; this is known as load shedding. Sometimes rejecting a request is the best way to stop an important system from becoming overwhelmed and being a bottleneck for multiple upstream services. Isolation The more one service depends on another being up, the more the health of one impacts the ability of the other to do its job. If we can use integration techniques that allow a downstream server to be offline, upstream services are less likely to be affected by outages, planned or unplanned.

Index A acceptance testing, Types of Tests access by reference, Access by Reference accountability, People adaptability, Summary Aegisthus project, Backup Data Pump aggregated logs, Logs, Logs, and Yet More Logs… antifragile systems, Microservices, The Antifragile Organization-Isolationbulkheads, Bulkheads circuit breakers, Circuit Breakers examples of, The Antifragile Organization increased use of, Microservices isolation, Isolation load shedding, Bulkheads timeouts, Timeouts AP systemdefinition of term, Sacrificing Consistency vs. CP system, AP or CP? API key-based authentication, API Keys, It’s All About the Keys application containers, Application Containers architects (see systems architects) architectural principlesdevelopment of, Principles Heroku's 12 factors, Principles key microservices principles, Bringing It All Together real-world example, A Real-World Example architectural safety, Architectural Safety, Architectural Safety Measures artifactsimages, Images as Artifacts operating system, Operating System Artifacts platform-specific, Platform-Specific Artifacts asynchronous collaborationcomplexities of, Complexities of Asynchronous Architectures implementing, Implementing Asynchronous Event-Based Collaboration vs. synchronous, Synchronous Versus Asynchronous ATOM specification, Technology Choices authentication/authorization, Authentication and Authorization-The Deputy Problemdefinition of terms, Authentication and Authorization fine-grained, Fine-Grained Authorization service-to-service, Service-to-Service Authentication and Authorization single sign-on (SSO), Common Single Sign-On Implementations single sign-on gateway, Single Sign-On Gateway terminology, Common Single Sign-On Implementations automationbenefits for deployment, Automation case studies on, Two Case Studies on the Power of Automation autonomymicroservices and, Autonomous role of systems architect in, Summary autoscaling, Autoscaling availabilityin CAP theorem, CAP Theorem key microservices principle of, How Much Is Too Much?

JSON web tokens (JWT), HMAC Over HTTP K Karyon, Tailored Service Template key-based authentication, API Keys Kibana, Logs, Logs, and Yet More Logs… L latency, How Much Is Too Much? Latency Monkey, The Antifragile Organization layered architectures, Microservices librariesclient, Client Libraries service metrics, Service Metrics shared, Shared Libraries Linux containers, Linux Containers load balancing, Load Balancing load shedding, Bulkheads local calls, Local Calls Are Not Like Remote Calls logsaggregated, Logs, Logs, and Yet More Logs…(see also monitoring) security issues, Logging standardization of, Standardization logstash, Logs, Logs, and Yet More Logs… loose coupling, Loose Coupling, Orchestration Versus Choreography, Loose and Tightly Coupled Organizations M man-in-the-middle attacks, Allow Everything Inside the Perimeter Marick's quadrant, Types of Tests maturity, Maturity mean time between failures (MTBF), Mean Time to Repair Over Mean Time Between Failures?

pages: 777 words: 186,993

Imagining India by Nandan Nilekani

Amazon: amazon.comamazon.co.ukamazon.deamazon.fr

affirmative action, Airbus A320, BRICs, British Empire, business process, business process outsourcing, call centre, clean water, colonial rule, corporate governance, cuban missile crisis, deindustrialization, demographic dividend, demographic transition, Deng Xiaoping, digital map, distributed generation, farmers can use mobile phones to check market prices, full employment, ghettoisation, glass ceiling, global supply chain, Hernando de Soto, income inequality, informal economy, Intergovernmental Panel on Climate Change (IPCC), joint-stock company, knowledge economy, labour market flexibility, land reform, light touch regulation, LNG terminal, load shedding, Mahatma Gandhi, market fragmentation, mass immigration, Mikhail Gorbachev, Network effects, new economy, New Urbanism, open economy, Parag Khanna, pension reform, Potemkin village, price mechanism, race to the bottom, rent control, rolodex, Ronald Reagan, school vouchers, Silicon Valley, smart grid, special economic zone, The Wealth of Nations by Adam Smith, Thomas L Friedman, Thomas Malthus, transaction costs, trickle-down economics, unemployed young men, upwardly mobile, urban planning, urban renewal, women in the workforce, working poor, working-age population

But I think that India’s growth rates have made us perhaps overly cheerful in the face of some obvious weaknesses. And somebody has to say it.” He is right, of course. There is no escaping our infrastructure problems, even here in India’s capital city—the Delhi newspapers during my visit have been full of headlines on the city’s power outages. A single day that week, as one outraged journalist wrote, had seen “10 periods of ‘load-shedding.’” In fact this particular bit of bureaucratspeak we use, “load-shedding,” reveals how a growing economy has found it difficult to look its crisis in the face.bt Our bad roads and power cuts are a reminder of our prereform years—it is here that we can most clearly see the evidence of India’s old structures, the tattered vestiges of socialism in an emerging free-market economy. As a result India now presents us with a bewildering landscape—of vibrant, private enterprise choking up as it meets crumbling public infrastructure.

bq The richer classes in Bombay’s slums earn as much as Rs 14,000 a month, which means a substantial part of Bombay’s middle class lives here. br It is a different story that this rule has often been violated by state governments. bs Bangalore can lay claim to pioneering bond issues by cities in India—India’s first such bond issue was by the Bangalore Mahanagarapalika in 1997 of Rs 1 billion, with a coupon rate of 13 percent. bt As a technical term, “load-shedding” means power cuts to tackle spikes in excess demand. But what India faces is consistent and severe power shortage. bu This view of roads and railways as an investment toward safety—to move people and goods in and out quickly, and avoid being cornered by enemies—has plenty of precedent. The Romans built Britain’s major road systems when they had occupied the restive island, and many of these still exist.

pages: 402 words: 110,972

Nerds on Wall Street: Math, Machines and Wired Markets by David J. Leinweber

Amazon: amazon.comamazon.co.ukamazon.deamazon.fr

AI winter, algorithmic trading, asset allocation, banking crisis, barriers to entry, Big bang: deregulation of the City of London, butterfly effect, buttonwood tree, buy low sell high, capital asset pricing model, citizen journalism, collateralized debt obligation, corporate governance, Craig Reynolds: boids flock, creative destruction, credit crunch, Credit Default Swap, credit default swaps / collateralized debt obligations, Danny Hillis, demand response, disintermediation, distributed generation, diversification, diversified portfolio, Emanuel Derman, en.wikipedia.org, experimental economics, financial innovation, fixed income, Gordon Gekko, implied volatility, index arbitrage, index fund, information retrieval, intangible asset, Internet Archive, John Nash: game theory, Kenneth Arrow, Khan Academy, load shedding, Long Term Capital Management, Machine translation of "The spirit is willing, but the flesh is weak." to Russian and back, market fragmentation, market microstructure, Mars Rover, Metcalfe’s law, moral hazard, mutually assured destruction, Myron Scholes, natural language processing, negative equity, Network effects, optical character recognition, paper trading, passive investing, pez dispenser, phenotype, prediction markets, quantitative hedge fund, quantitative trading / quantitative finance, QWERTY keyboard, RAND corporation, random walk, Ray Kurzweil, Renaissance Technologies, Richard Stallman, risk tolerance, risk-adjusted returns, risk/return, Robert Metcalfe, Ronald Reagan, Rubik’s Cube, semantic web, Sharpe ratio, short selling, Silicon Valley, Small Order Execution System, smart grid, smart meter, social web, South Sea Bubble, statistical arbitrage, statistical model, Steve Jobs, Steven Levy, Tacoma Narrows Bridge, the scientific method, The Wisdom of Crowds, time value of money, too big to fail, transaction costs, Turing machine, Upton Sinclair, value at risk, Vernor Vinge, yield curve, Yogi Berra, your tax dollars at work

Smart meters with Internet communication capabilities were the enabling technology for the next wave of technologydriven savings, designed to bring the suppliers into the process. This was 336 Nerds on Wall Str eet done using simple command-and-control load-shedding measures that allowed utilities, with prior agreement of larger customers, to shed loads during periods of peak demand or to remotely adjust airconditioning thermostats upward to reduce demand when needed. The next step, still in its infancy, is to introduce real-time electricity pricing. This will allow consumers to make their own economically motivated load-shedding decisions, and to program their meters to implement those decisions for them. Day Trading for Electrons In the best seller Hot, Flat, and Crowded: Why We Need a Green Revolution—and How It Can Renew America (Farrar, Straus & Giroux, 2008), New York Times columnist Thomas Friedman calls the use of smart meters to create a market-based system of energy technology (ET), “when IT meets ET—day trading for electrons.”

pages: 741 words: 179,454

Extreme Money: Masters of the Universe and the Cult of Risk by Satyajit Das

Amazon: amazon.comamazon.co.ukamazon.deamazon.fr

affirmative action, Albert Einstein, algorithmic trading, Andy Kessler, Asian financial crisis, asset allocation, asset-backed security, bank run, banking crisis, banks create money, Basel III, Benoit Mandelbrot, Berlin Wall, Bernie Madoff, Big bang: deregulation of the City of London, Black Swan, Bonfire of the Vanities, bonus culture, Bretton Woods, BRICs, British Empire, capital asset pricing model, Carmen Reinhart, carried interest, Celtic Tiger, clean water, cognitive dissonance, collapse of Lehman Brothers, collateralized debt obligation, corporate governance, corporate raider, creative destruction, credit crunch, Credit Default Swap, credit default swaps / collateralized debt obligations, Daniel Kahneman / Amos Tversky, debt deflation, Deng Xiaoping, deskilling, discrete time, diversification, diversified portfolio, Doomsday Clock, Edward Thorp, Emanuel Derman, en.wikipedia.org, Eugene Fama: efficient market hypothesis, eurozone crisis, Fall of the Berlin Wall, financial independence, financial innovation, financial thriller, fixed income, full employment, global reserve currency, Goldman Sachs: Vampire Squid, Gordon Gekko, greed is good, happiness index / gross national happiness, haute cuisine, high net worth, Hyman Minsky, index fund, information asymmetry, interest rate swap, invention of the wheel, invisible hand, Isaac Newton, job automation, Johann Wolfgang von Goethe, John Meriwether, joint-stock company, Joseph Schumpeter, Kenneth Arrow, Kenneth Rogoff, Kevin Kelly, labour market flexibility, laissez-faire capitalism, load shedding, locking in a profit, Long Term Capital Management, Louis Bachelier, margin call, market bubble, market fundamentalism, Marshall McLuhan, Martin Wolf, mega-rich, merger arbitrage, Mikhail Gorbachev, Milgram experiment, money market fund, Mont Pelerin Society, moral hazard, mortgage debt, mortgage tax deduction, mutually assured destruction, Myron Scholes, Naomi Klein, negative equity, Network effects, new economy, Nick Leeson, Nixon shock, Northern Rock, nuclear winter, oil shock, Own Your Own Home, Paul Samuelson, pets.com, Philip Mirowski, Plutocrats, plutocrats, Ponzi scheme, price anchoring, price stability, profit maximization, quantitative easing, quantitative trading / quantitative finance, Ralph Nader, RAND corporation, random walk, Ray Kurzweil, regulatory arbitrage, rent control, rent-seeking, reserve currency, Richard Feynman, Richard Feynman, Richard Thaler, Right to Buy, risk-adjusted returns, risk/return, road to serfdom, Robert Shiller, Robert Shiller, Rod Stewart played at Stephen Schwarzman birthday party, rolodex, Ronald Reagan, Ronald Reagan: Tear down this wall, Satyajit Das, savings glut, shareholder value, Sharpe ratio, short selling, Silicon Valley, six sigma, Slavoj Žižek, South Sea Bubble, special economic zone, statistical model, Stephen Hawking, Steve Jobs, survivorship bias, The Chicago School, The Great Moderation, the market place, the medium is the message, The Myth of the Rational Market, The Nature of the Firm, the new new thing, The Predators' Ball, The Wealth of Nations by Adam Smith, Thorstein Veblen, too big to fail, trickle-down economics, Turing test, Upton Sinclair, value at risk, Yogi Berra, zero-coupon bond, zero-sum game

My Indian heritage has gotten me an invitation to the Indian Bankers’ Association party themed “India Shining,” a shameless sales pitch for investment in India. A minister extols the virtues of India, citing statistics on growth, resource availability, and opportunities. There is no mention of the fact that the vast majority of the Indian population has no access to sanitation, clean water, education, or healthcare. There is no mention of the aging colonial era infrastructure where inadequate electricity supply results in daily load shedding or brownouts interrupting power supplies for several hours most days. An American banker finds India fascinating and full of opportunity. “A billion people, a billion consumers, wow!” He is fascinated that I, an Indian, do not speak Indian but Bengali, one of the hundreds of languages spoken in India. “Wouldn’t it be easier if everybody just spoke, you know, Indian?” Dan Quayle, a former U.S. vice-president, once apologized to Latin Americans that he could not speak Latin.

pages: 834 words: 180,700

The Architecture of Open Source Applications by Amy Brown, Greg Wilson

Amazon: amazon.comamazon.co.ukamazon.deamazon.fr

8-hour work day, anti-pattern, bioinformatics, c2.com, cloud computing, collaborative editing, combinatorial explosion, computer vision, continuous integration, create, read, update, delete, David Heinemeier Hansson, Debian, domain-specific language, Donald Knuth, en.wikipedia.org, fault tolerance, finite state, Firefox, friendly fire, Guido van Rossum, linked data, load shedding, locality of reference, loose coupling, Mars Rover, MVC pattern, peer-to-peer, Perl 6, premature optimization, recommendation engine, revision control, Ruby on Rails, side project, Skype, slashdot, social web, speech recognition, the scientific method, The Wisdom of Crowds, web application, WebSocket

Like the consistent hash ring approach, this range partitioning splits the keyspace into ranges, with each key range being managed by one machine and potentially replicated to others. Unlike the consistent hashing approach, two keys that are next to each other in the key's sort order are likely to appear in the same partition. This reduces the size of the routing metadata, as large ranges are compressed to [start, end] markers. In adding active record-keeping of the range-to-server mapping, the range partitioning approach allows for more fine-grained control of load-shedding from heavily loaded servers. If a specific key range sees higher traffic than other ranges, a load manager can reduce the size of the range on that server, or reduce the number of shards that this server serves. The added freedom to actively manage load comes at the expense of extra architectural components which monitor and route shards. The BigTable Way Google's BigTable paper describes a range-partitioning hierarchical technique for sharding data into tablets.