Designing Data-Intensive Applications PDF

Designing Data-Intensive Applications THE BIG IDEAS BEHIND RELIABLE, SCALABLE, AND MAINTAINABLE SYSTEMS Martin Kleppmann  Designing Data-Intensive Applications The Big Ideas Behind Reliable, Scalable, and Maintainable Syst...

Designing Data-Intensive Applications THE BIG IDEAS BEHIND RELIABLE, SCALABLE, AND MAINTAINABLE SYSTEMS Martin Kleppmann  Designing Data-Intensive Applications The Big Ideas Behind Reliable, Scalable, and Maintainable Systems Martin Kleppmann Beijing Boston Farnham Sebastopol Tokyo Designing Data-Intensive Applications by Martin Kleppmann Copyright © 2017 Martin Kleppmann. All rights reserved. Printed in the United States of America. Published by O’Reilly Media, Inc., 1005 Gravenstein Highway North, Sebastopol, CA 95472. O’Reilly books may be purchased for educational, business, or sales promotional use. Online editions are also available for most titles (http://oreilly.com/safari). For more information, contact our corporate/insti‐ tutional sales department: 800-998-9938 or [email protected]. Editors: Ann Spencer and Marie Beaugureau Indexer: Ellen Troutman-Zaig Production Editor: Kristen Brown Interior Designer: David Futato Copyeditor: Rachel Head Cover Designer: Karen Montgomery Proofreader: Amanda Kersey Illustrator: Rebecca Demarest March 2017: First Edition Revision History for the First Edition 2017-03-01: First Release See http://oreilly.com/catalog/errata.csp?isbn=9781449373320 for release details. The O’Reilly logo is a registered trademark of O’Reilly Media, Inc. Designing Data-Intensive Applications, the cover image, and related trade dress are trademarks of O’Reilly Media, Inc. While the publisher and the author have used good faith efforts to ensure that the information and instructions contained in this work are accurate, the publisher and the author disclaim all responsibility for errors or omissions, including without limitation responsibility for damages resulting from the use of or reliance on this work. Use of the information and instructions contained in this work is at your own risk. If any code samples or other technology this work contains or describes is subject to open source licenses or the intellectual property rights of others, it is your responsibility to ensure that your use thereof complies with such licenses and/or rights. 978-1-449-37332-0 [LSI]  Technology is a powerful force in our society. Data, software, and communication can be used for bad: to entrench unfair power structures, to undermine human rights, and to protect vested interests. But they can also be used for good: to make underrepresented people’s voices heard, to create opportunities for everyone, and to avert disasters. This book is dedicated to everyone working toward the good. Computing is pop culture. […] Pop culture holds a disdain for history. Pop culture is all about identity and feeling like you’re participating. It has nothing to do with cooperation, the past or the future—it’s living in the present. I think the same is true of most people who write code for money. They have no idea where [their culture came from]. —Alan Kay, in interview with Dr Dobb’s Journal (2012)  Table of Contents Preface...................................................................... xiii Part I. Foundations of Data Systems 1. Reliable, Scalable, and Maintainable Applications............................... 3 Thinking About Data Systems 4 Reliability 6 Hardware Faults 7 Software Errors 8 Human Errors 9 How Important Is Reliability? 10 Scalability 10 Describing Load 11 Describing Performance 13 Approaches for Coping with Load 17 Maintainability 18 Operability: Making Life Easy for Operations 19 Simplicity: Managing Complexity 20 Evolvability: Making Change Easy 21 Summary 22 2. Data Models and Query Languages........................................... 27 Relational Model Versus Document Model 28 The Birth of NoSQL 29 The Object-Relational Mismatch 29 Many-to-One and Many-to-Many Relationships 33 Are Document Databases Repeating History? 36 vii  Relational Versus Document Databases Today 38 Query Languages for Data 42 Declarative Queries on the Web 44 MapReduce Querying 46 Graph-Like Data Models 49 Property Graphs 50 The Cypher Query Language 52 Graph Queries in SQL 53 Triple-Stores and SPARQL 55 The Foundation: Datalog 60 Summary 63 3. Storage and Retrieval...................................................... 69 Data Structures That Power Your Database 70 Hash Indexes 72 SSTables and LSM-Trees 76 B-Trees 79 Comparing B-Trees and LSM-Trees 83 Other Indexing Structures 85 Transaction Processing or Analytics? 90 Data Warehousing 91 Stars and Snowflakes: Schemas for Analytics 93 Column-Oriented Storage 95 Column Compression 97 Sort Order in Column Storage 99 Writing to Column-Oriented Storage 101 Aggregation: Data Cubes and Materialized Views 101 Summary 103 4. Encoding and Evolution................................................... 111 Formats for Encoding Data 112 Language-Specific Formats 113 JSON, XML, and Binary Variants 114 Thrift and Protocol Buffers 117 Avro 122 The Merits of Schemas 127 Modes of Dataflow 128 Dataflow Through Databases 129 Dataflow Through Services: REST and RPC 131 Message-Passing Dataflow 136 Summary 139 viii | Table of Contents Part II. Distributed Data 5. Replication.............................................................. 151 Leaders and Followers 152 Synchronous Versus Asynchronous Replication 153 Setting Up New Followers 155 Handling Node Outages 156 Implementation of Replication Logs 158 Problems with Replication Lag 161 Reading Your Own Writes 162 Monotonic Reads 164 Consistent Prefix Reads 165 Solutions for Replication Lag 167 Multi-Leader Replication 168 Use Cases for Multi-Leader Replication 168 Handling Write Conflicts 171 Multi-Leader Replication Topologies 175 Leaderless Replication 177 Writing to the Database When a Node Is Down 177 Limitations of Quorum Consistency 181 Sloppy Quorums and Hinted Handoff 183 Detecting Concurrent Writes 184 Summary 192 6. Partitioning............................................................. 199 Partitioning and Replication 200 Partitioning of Key-Value Data 201 Partitioning by Key Range 202 Partitioning by Hash of Key 203 Skewed Workloads and Relieving Hot Spots 205 Partitioning and Secondary Indexes 206 Partitioning Secondary Indexes by Document 206 Partitioning Secondary Indexes by Term 208 Rebalancing Partitions 209 Strategies for Rebalancing 210 Operations: Automatic or Manual Rebalancing 213 Request Routing 214 Parallel Query Execution 216 Summary 216 7. Transactions............................................................. 221 The Slippery Concept of a Transaction 222 Table of Contents | ix  The Meaning of ACID 223 Single-Object and Multi-Object Operations 228 Weak Isolation Levels 233 Read Committed 234 Snapshot Isolation and Repeatable Read 237 Preventing Lost Updates 242 Write Skew and Phantoms 246 Serializability 251 Actual Serial Execution 252 Two-Phase Locking (2PL) 257 Serializable Snapshot Isolation (SSI) 261 Summary 266 8. The Trouble with Distributed Systems....................................... 273 Faults and Partial Failures 274 Cloud Computing and Supercomputing 275 Unreliable Networks 277 Network Faults in Practice 279 Detecting Faults 280 Timeouts and Unbounded Delays 281 Synchronous Versus Asynchronous Networks 284 Unreliable Clocks 287 Monotonic Versus Time-of-Day Clocks 288 Clock Synchronization and Accuracy 289 Relying on Synchronized Clocks 291 Process Pauses 295 Knowledge, Truth, and Lies 300 The Truth Is Defined by the Majority 300 Byzantine Faults 304 System Model and Reality 306 Summary 310 9. Consistency and Consensus................................................ 321 Consistency Guarantees 322 Linearizability 324 What Makes a System Linearizable? 325 Relying on Linearizability 330 Implementing Linearizable Systems 332 The Cost of Linearizability 335 Ordering Guarantees 339 Ordering and Causality 339 Sequence Number Ordering 343 x | Table of Contents  Total Order Broadcast 348 Distributed Transactions and Consensus 352 Atomic Commit and Two-Phase Commit (2PC) 354 Distributed Transactions in Practice 360 Fault-Tolerant Consensus 364 Membership and Coordination Services 370 Summary 373 Part III. Derived Data 10. Batch Processing......................................................... 389 Batch Processing with Unix Tools 391 Simple Log Analysis 391 The Unix Philosophy 394 MapReduce and Distributed Filesystems 397 MapReduce Job Execution 399 Reduce-Side Joins and Grouping 403 Map-Side Joins 408 The Output of Batch Workflows 411 Comparing Hadoop to Distributed Databases 414 Beyond MapReduce 419 Materialization of Intermediate State 419 Graphs and Iterative Processing 424 High-Level APIs and Languages 426 Summary 429 11. Stream Processing........................................................ 439 Transmitting Event Streams 440 Messaging Systems 441 Partitioned Logs 446 Databases and Streams 451 Keeping Systems in Sync 452 Change Data Capture 454 Event Sourcing 457 State, Streams, and Immutability 459 Processing Streams 464 Uses of Stream Processing 465 Reasoning About Time 468 Stream Joins 472 Fault Tolerance 476 Summary 479 Table of Contents | xi 12. The Future of Data Systems................................................ 489 Data Integration 490 Combining Specialized Tools by Deriving Data 490 Batch and Stream Processing 494 Unbundling Databases 499 Composing Data Storage Technologies 499 Designing Applications Around Dataflow 504 Observing Derived State 509 Aiming for Correctness 515 The End-to-End Argument for Databases 516 Enforcing Constraints 521 Timeliness and Integrity 524 Trust, but Verify 528 Doing the Right Thing 533 Predictive Analytics 533 Privacy and Tracking 536 Summary 543 Glossary..................................................................... 553 Index....................................................................... 559 xii | Table of Contents  Preface If you have worked in software engineering in recent years, especially in server-side and backend systems, you have probably been bombarded with a plethora of buzz‐ words relating to storage and processing of data. NoSQL! Big Data! Web-scale! Sharding! Eventual consistency! ACID! CAP theorem! Cloud services! MapReduce! Real-time! In the last decade we have seen many interesting developments in databases, in dis‐ tributed systems, and in the ways we build applications on top of them. There are various driving forces for these developments: Internet companies such as Google, Yahoo!, Amazon, Facebook, LinkedIn, Microsoft, and Twitter are handling huge volumes of data and traffic, forcing them to create new tools that enable them to efficiently handle such scale. Businesses need to be agile, test hypotheses cheaply, and respond quickly to new market insights by keeping development cycles short and data models flexible. Free and open source software has become very successful and is now preferred to commercial or bespoke in-house software in many environments. CPU clock speeds are barely increasing, but multi-core processors are standard, and networks are getting faster. This means parallelism is only going to increase. Even if you work on a small team, you can now build systems that are distributed across many machines and even multiple geographic regions, thanks to infra‐ structure as a service (IaaS) such as Amazon Web Services. Many services are now expected to be highly available; extended downtime due to outages or maintenance is becoming increasingly unacceptable. Data-intensive applications are pushing the boundaries of what is possible by making use of these technological developments. We call an application data-intensive if data is its primary challenge—the quantity of data, the complexity of data, or the speed at Preface | xiii which it is changing—as opposed to compute-intensive, where CPU cycles are the bottleneck. The tools and technologies that help data-intensive applications store and process data have been rapidly adapting to these changes. New types of database systems (“NoSQL”) have been getting lots of attention, but message queues, caches, search indexes, frameworks for batch and stream processing, and related technologies are very important too. Many applications use some combination of these. The buzzwords that fill this space are a sign of enthusiasm for the new possibilities, which is a great thing. However, as software engineers and architects, we also need to have a technically accurate and precise understanding of the various technologies and their trade-offs if we want to build good applications. For that understanding, we have to dig deeper than buzzwords. Fortunately, behind the rapid changes in technology, there are enduring principles that remain true, no matter which version of a particular tool you are using. If you understand those principles, you’re in a position to see where each tool fits in, how to make good use of it, and how to avoid its pitfalls. That’s where this book comes in. The goal of this book is to help you navigate the diverse and fast-changing landscape of technologies for processing and storing data. This book is not a tutorial for one particular tool, nor is it a textbook full of dry theory. Instead, we will look at examples of successful data systems: technologies that form the foundation of many popular applications and that have to meet scalability, performance, and reliability require‐ ments in production every day. We will dig into the internals of those systems, tease apart their key algorithms, dis‐ cuss their principles and the trade-offs they have to make. On this journey, we will try to find useful ways of thinking about data systems—not just how they work, but also why they work that way, and what questions we need to ask. After reading this book, you will be in a great position to decide which kind of tech‐ nology is appropriate for which purpose, and understand how tools can be combined to form the foundation of a good application architecture. You won’t be ready to build your own database storage engine from scratch, but fortunately that is rarely necessary. You will, however, develop a good intuition for what your systems are doing under the hood so that you can reason about their behavior, make good design decisions, and track down any problems that may arise. Who Should Read This Book? If you develop applications that have some kind of server/backend for storing or pro‐ cessing data, and your applications use the internet (e.g., web applications, mobile apps, or internet-connected sensors), then this book is for you. xiv | Preface This book is for software engineers, software architects, and technical managers who love to code. It is especially relevant if you need to make decisions about the architec‐ ture of the systems you work on—for example, if you need to choose tools for solving a given problem and figure out how best to apply them. But even if you have no choice over your tools, this book will help you better understand their strengths and weaknesses. You should have some experience building web-based applications or network serv‐ ices, and you should be familiar with relational databases and SQL. Any non- relational databases and other data-related tools you know are a bonus, but not required. A general understanding of common network protocols like TCP and HTTP is helpful. Your choice of programming language or framework makes no dif‐ ference for this book. If any of the following are true for you, you’ll find this book valuable: You want to learn how to make data systems scalable, for example, to support web or mobile apps with millions of users. You need to make applications highly available (minimizing downtime) and operationally robust. You are looking for ways of making systems easier to maintain in the long run, even as they grow and as requirements and technologies change. You have a natural curiosity for the way things work and want to know what goes on inside major websites and online services. This book breaks down the internals of various databases and data processing systems, and it’s great fun to explore the bright thinking that went into their design. Sometimes, when discussing scalable data systems, people make comments along the lines of, “You’re not Google or Amazon. Stop worrying about scale and just use a relational database.” There is truth in that statement: building for scale that you don’t need is wasted effort and may lock you into an inflexible design. In effect, it is a form of premature optimization. However, it’s also important to choose the right tool for the job, and different technologies each have their own strengths and weaknesses. As we shall see, relational databases are important but not the final word on dealing with data. Scope of This Book This book does not attempt to give detailed instructions on how to install or use spe‐ cific software packages or APIs, since there is already plenty of documentation for those things. Instead we discuss the various principles and trade-offs that are funda‐ mental to data systems, and we explore the different design decisions taken by differ‐ ent products. Preface | xv In the ebook editions we have included links to the full text of online resources. All links were verified at the time of publication, but unfortunately links tend to break frequently due to the nature of the web. If you come across a broken link, or if you are reading a print copy of this book, you can look up references using a search engine. For academic papers, you can search for the title in Google Scholar to find open-access PDF files. Alternatively, you can find all of the references at https:// github.com/ept/ddia-references, where we maintain up-to-date links. We look primarily at the architecture of data systems and the ways they are integrated into data-intensive applications. This book doesn’t have space to cover deployment, operations, security, management, and other areas—those are complex and impor‐ tant topics, and we wouldn’t do them justice by making them superficial side notes in this book. They deserve books of their own. Many of the technologies described in this book fall within the realm of the Big Data buzzword. However, the term “Big Data” is so overused and underdefined that it is not useful in a serious engineering discussion. This book uses less ambiguous terms, such as single-node versus distributed systems, or online/interactive versus offline/ batch processing systems. This book has a bias toward free and open source software (FOSS), because reading, modifying, and executing source code is a great way to understand how something works in detail. Open platforms also reduce the risk of vendor lock-in. However, where appropriate, we also discuss proprietary software (closed-source software, soft‐ ware as a service, or companies’ in-house software that is only described in literature but not released publicly). Outline of This Book This book is arranged into three parts: 1. In Part I, we discuss the fundamental ideas that underpin the design of data- intensive applications. We start in Chapter 1 by discussing what we’re actually trying to achieve: reliability, scalability, and maintainability; how we need to think about them; and how we can achieve them. In Chapter 2 we compare sev‐ eral different data models and query languages, and see how they are appropriate to different situations. In Chapter 3 we talk about storage engines: how databases arrange data on disk so that we can find it again efficiently. Chapter 4 turns to formats for data encoding (serialization) and evolution of schemas over time. 2. In Part II, we move from data stored on one machine to data that is distributed across multiple machines. This is often necessary for scalability, but brings with it a variety of unique challenges. We first discuss replication (Chapter 5), parti‐ tioning/sharding (Chapter 6), and transactions (Chapter 7). We then go into xvi | Preface  more detail on the problems with distributed systems (Chapter 8) and what it means to achieve consistency and consensus in a distributed system (Chapter 9). 3. In Part III, we discuss systems that derive some datasets from other datasets. Derived data often occurs in heterogeneous systems: when there is no one data‐ base that can do everything well, applications need to integrate several different databases, caches, indexes, and so on. In Chapter 10 we start with a batch pro‐ cessing approach to derived data, and we build upon it with stream processing in Chapter 11. Finally, in Chapter 12 we put everything together and discuss approaches for building reliable, scalable, and maintainable applications in the future. References and Further Reading Most of what we discuss in this book has already been said elsewhere in some form or another—in conference presentations, research papers, blog posts, code, bug trackers, mailing lists, and engineering folklore. This book summarizes the most important ideas from many different sources, and it includes pointers to the original literature throughout the text. The references at the end of each chapter are a great resource if you want to explore an area in more depth, and most of them are freely available online. O’Reilly Safari Safari (formerly Safari Books Online) is a membership-based training and reference platform for enterprise, government, educators, and individuals. Members have access to thousands of books, training videos, Learning Paths, interac‐ tive tutorials, and curated playlists from over 250 publishers, including O’Reilly Media, Harvard Business Review, Prentice Hall Professional, Addison-Wesley Pro‐ fessional, Microsoft Press, Sams, Que, Peachpit Press, Adobe, Focal Press, Cisco Press, John Wiley & Sons, Syngress, Morgan Kaufmann, IBM Redbooks, Packt, Adobe Press, FT Press, Apress, Manning, New Riders, McGraw-Hill, Jones & Bartlett, and Course Technology, among others. For more information, please visit http://oreilly.com/safari. Preface | xvii How to Contact Us Please address comments and questions concerning this book to the publisher: O’Reilly Media, Inc. 1005 Gravenstein Highway North Sebastopol, CA 95472 800-998-9938 (in the United States or Canada) 707-829-0515 (international or local) 707-829-0104 (fax) We have a web page for this book, where we list errata, examples, and any additional information. You can access this page at http://bit.ly/designing-data-intensive-apps. To comment or ask technical questions about this book, send email to bookques‐ [email protected]. For more information about our books, courses, conferences, and news, see our web‐ site at http://www.oreilly.com. Find us on Facebook: http://facebook.com/oreilly Follow us on Twitter: http://twitter.com/oreillymedia Watch us on YouTube: http://www.youtube.com/oreillymedia Acknowledgments This book is an amalgamation and systematization of a large number of other peo‐ ple’s ideas and knowledge, combining experience from both academic research and industrial practice. In computing we tend to be attracted to things that are new and shiny, but I think we have a huge amount to learn from things that have been done before. This book has over 800 references to articles, blog posts, talks, documenta‐ tion, and more, and they have been an invaluable learning resource for me. I am very grateful to the authors of this material for sharing their knowledge. I have also learned a lot from personal conversations, thanks to a large number of people who have taken the time to discuss ideas or patiently explain things to me. In particular, I would like to thank Joe Adler, Ross Anderson, Peter Bailis, Márton Balassi, Alastair Beresford, Mark Callaghan, Mat Clayton, Patrick Collison, Sean Cribbs, Shirshanka Das, Niklas Ekström, Stephan Ewen, Alan Fekete, Gyula Fóra, Camille Fournier, Andres Freund, John Garbutt, Seth Gilbert, Tom Haggett, Pat Hel‐ land, Joe Hellerstein, Jakob Homan, Heidi Howard, John Hugg, Julian Hyde, Conrad Irwin, Evan Jones, Flavio Junqueira, Jessica Kerr, Kyle Kingsbury, Jay Kreps, Carl Lerche, Nicolas Liochon, Steve Loughran, Lee Mallabone, Nathan Marz, Caitie xviii | Preface McCaffrey, Josie McLellan, Christopher Meiklejohn, Ian Meyers, Neha Narkhede, Neha Narula, Cathy O’Neil, Onora O’Neill, Ludovic Orban, Zoran Perkov, Julia Powles, Chris Riccomini, Henry Robinson, David Rosenthal, Jennifer Rullmann, Matthew Sackman, Martin Scholl, Amit Sela, Gwen Shapira, Greg Spurrier, Sam Stokes, Ben Stopford, Tom Stuart, Diana Vasile, Rahul Vohra, Pete Warden, and Brett Wooldridge. Several more people have been invaluable to the writing of this book by reviewing drafts and providing feedback. For these contributions I am particularly indebted to Raul Agepati, Tyler Akidau, Mattias Andersson, Sasha Baranov, Veena Basavaraj, David Beyer, Jim Brikman, Paul Carey, Raul Castro Fernandez, Joseph Chow, Derek Elkins, Sam Elliott, Alexander Gallego, Mark Grover, Stu Halloway, Heidi Howard, Nicola Kleppmann, Stefan Kruppa, Bjorn Madsen, Sander Mak, Stefan Podkowinski, Phil Potter, Hamid Ramazani, Sam Stokes, and Ben Summers. Of course, I take all responsibility for any remaining errors or unpalatable opinions in this book. For helping this book become real, and for their patience with my slow writing and unusual requests, I am grateful to my editors Marie Beaugureau, Mike Loukides, Ann Spencer, and all the team at O’Reilly. For helping find the right words, I thank Rachel Head. For giving me the time and freedom to write in spite of other work commit‐ ments, I thank Alastair Beresford, Susan Goodhue, Neha Narkhede, and Kevin Scott. Very special thanks are due to Shabbir Diwan and Edie Freedman, who illustrated with great care the maps that accompany the chapters. It’s wonderful that they took on the unconventional idea of creating maps, and made them so beautiful and com‐ pelling. Finally, my love goes to my family and friends, without whom I would not have been able to get through this writing process that has taken almost four years. You’re the best. Preface | xix  PART I Foundations of Data Systems The first four chapters go through the fundamental ideas that apply to all data sys‐ tems, whether running on a single machine or distributed across a cluster of machines: 1. Chapter 1 introduces the terminology and approach that we’re going to use throughout this book. It examines what we actually mean by words like reliabil‐ ity, scalability, and maintainability, and how we can try to achieve these goals. 2. Chapter 2 compares several different data models and query languages—the most visible distinguishing factor between databases from a developer’s point of view. We will see how different models are appropriate to different situations. 3. Chapter 3 turns to the internals of storage engines and looks at how databases lay out data on disk. Different storage engines are optimized for different workloads, and choosing the right one can have a huge effect on performance. 4. Chapter 4 compares various formats for data encoding (serialization) and espe‐ cially examines how they fare in an environment where application requirements change and schemas need to adapt over time. Later, Part II will turn to the particular issues of distributed data systems.  CHAPTER 1 Reliable, Scalable, and Maintainable Applications The Internet was done so well that most people think of it as a natural resource like the Pacific Ocean, rather than something that was man-made. When was the last time a tech‐ nology with a scale like that was so error-free? —Alan Kay, in interview with Dr Dobb’s Journal (2012) Many applications today are data-intensive, as opposed to compute-intensive. Raw CPU power is rarely a limiting factor for these applications—bigger problems are usually the amount of data, the complexity of data, and the speed at which it is changing. A data-intensive application is typically built from standard building blocks that pro‐ vide commonly needed functionality. For example, many applications need to: Store data so that they, or another application, can find it again later (databases) Remember the result of an expensive operation, to speed up reads (caches) Allow users to search data by keyword or filter it in various ways (search indexes) Send a message to another process, to be handled asynchronously (stream pro‐ cessing) Periodically crunch a large amount of accumulated data (batch processing) If that sounds painfully obvious, that’s just because these data systems are such a suc‐ cessful abstraction: we use them all the time without thinking too much. When build‐ ing an application, most engineers wouldn’t dream of writing a new data storage engine from scratch, because databases are a perfectly good tool for the job. 3 But reality is not that simple. There are many database systems with different charac‐ teristics, because different applications have different requirements. There are vari‐ ous approaches to caching, several ways of building search indexes, and so on. When building an application, we still need to figure out which tools and which approaches are the most appropriate for the task at hand. And it can be hard to combine tools when you need to do something that a single tool cannot do alone. This book is a journey through both the principles and the practicalities of data sys‐ tems, and how you can use them to build data-intensive applications. We will explore what different tools have in common, what distinguishes them, and how they achieve their characteristics. In this chapter, we will start by exploring the fundamentals of what we are trying to achieve: reliable, scalable, and maintainable data systems. We’ll clarify what those things mean, outline some ways of thinking about them, and go over the basics that we will need for later chapters. In the following chapters we will continue layer by layer, looking at different design decisions that need to be considered when working on a data-intensive application. Thinking About Data Systems We typically think of databases, queues, caches, etc. as being very different categories of tools. Although a database and a message queue have some superficial similarity— both store data for some time—they have very different access patterns, which means different performance characteristics, and thus very different implementations. So why should we lump them all together under an umbrella term like data systems? Many new tools for data storage and processing have emerged in recent years. They are optimized for a variety of different use cases, and they no longer neatly fit into traditional categories. For example, there are datastores that are also used as mes‐ sage queues (Redis), and there are message queues with database-like durability guar‐ antees (Apache Kafka). The boundaries between the categories are becoming blurred. Secondly, increasingly many applications now have such demanding or wide-ranging requirements that a single tool can no longer meet all of its data processing and stor‐ age needs. Instead, the work is broken down into tasks that can be performed effi‐ ciently on a single tool, and those different tools are stitched together using application code. For example, if you have an application-managed caching layer (using Memcached or similar), or a full-text search server (such as Elasticsearch or Solr) separate from your main database, it is normally the application code’s responsibility to keep those caches and indexes in sync with the main database. Figure 1-1 gives a glimpse of what this may look like (we will go into detail in later chapters). 4 | Chapter 1: Reliable, Scalable, and Maintainable Applications Figure 1-1. One possible architecture for a data system that combines several components. When you combine several tools in order to provide a service, the service’s interface or application programming interface (API) usually hides those implementation details from clients. Now you have essentially created a new, special-purpose data system from smaller, general-purpose components. Your composite data system may provide certain guarantees: e.g., that the cache will be correctly invalidated or upda‐ ted on writes so that outside clients see consistent results. You are now not only an application developer, but also a data system designer. If you are designing a data system or service, a lot of tricky questions arise. How do you ensure that the data remains correct and complete, even when things go wrong internally? How do you provide consistently good performance to clients, even when parts of your system are degraded? How do you scale to handle an increase in load? What does a good API for the service look like? There are many factors that may influence the design of a data system, including the skills and experience of the people involved, legacy system dependencies, the time‐ scale for delivery, your organization’s tolerance of different kinds of risk, regulatory constraints, etc. Those factors depend very much on the situation. Thinking About Data Systems | 5 In this book, we focus on three concerns that are important in most software systems: Reliability The system should continue to work correctly (performing the correct function at the desired level of performance) even in the face of adversity (hardware or soft‐ ware faults, and even human error). See “Reliability” on page 6. Scalability As the system grows (in data volume, traffic volume, or complexity), there should be reasonable ways of dealing with that growth. See “Scalability” on page 10. Maintainability Over time, many different people will work on the system (engineering and oper‐ ations, both maintaining current behavior and adapting the system to new use cases), and they should all be able to work on it productively. See “Maintainabil‐ ity” on page 18. These words are often cast around without a clear understanding of what they mean. In the interest of thoughtful engineering, we will spend the rest of this chapter exploring ways of thinking about reliability, scalability, and maintainability. Then, in the following chapters, we will look at various techniques, architectures, and algo‐ rithms that are used in order to achieve those goals. Reliability Everybody has an intuitive idea of what it means for something to be reliable or unre‐ liable. For software, typical expectations include: The application performs the function that the user expected. It can tolerate the user making mistakes or using the software in unexpected ways. Its performance is good enough for the required use case, under the expected load and data volume. The system prevents any unauthorized access and abuse. If all those things together mean “working correctly,” then we can understand relia‐ bility as meaning, roughly, “continuing to work correctly, even when things go wrong.” The things that can go wrong are called faults, and systems that anticipate faults and can cope with them are called fault-tolerant or resilient. The former term is slightly misleading: it suggests that we could make a system tolerant of every possible kind of fault, which in reality is not feasible. If the entire planet Earth (and all servers on it) were swallowed by a black hole, tolerance of that fault would require web hosting in 6 | Chapter 1: Reliable, Scalable, and Maintainable Applications space—good luck getting that budget item approved. So it only makes sense to talk about tolerating certain types of faults. Note that a fault is not the same as a failure. A fault is usually defined as one com‐ ponent of the system deviating from its spec, whereas a failure is when the system as a whole stops providing the required service to the user. It is impossible to reduce the probability of a fault to zero; therefore it is usually best to design fault-tolerance mechanisms that prevent faults from causing failures. In this book we cover several techniques for building reliable systems from unreliable parts. Counterintuitively, in such fault-tolerant systems, it can make sense to increase the rate of faults by triggering them deliberately—for example, by randomly killing indi‐ vidual processes without warning. Many critical bugs are actually due to poor error handling ; by deliberately inducing faults, you ensure that the fault-tolerance machinery is continually exercised and tested, which can increase your confidence that faults will be handled correctly when they occur naturally. The Netflix Chaos Monkey is an example of this approach. Although we generally prefer tolerating faults over preventing faults, there are cases where prevention is better than cure (e.g., because no cure exists). This is the case with security matters, for example: if an attacker has compromised a system and gained access to sensitive data, that event cannot be undone. However, this book mostly deals with the kinds of faults that can be cured, as described in the following sections. Hardware Faults When we think of causes of system failure, hardware faults quickly come to mind. Hard disks crash, RAM becomes faulty, the power grid has a blackout, someone unplugs the wrong network cable. Anyone who has worked with large datacenters can tell you that these things happen all the time when you have a lot of machines. Hard disks are reported as having a mean time to failure (MTTF) of about 10 to 50 years [5, 6]. Thus, on a storage cluster with 10,000 disks, we should expect on average one disk to die per day. Our first response is usually to add redundancy to the individual hardware compo‐ nents in order to reduce the failure rate of the system. Disks may be set up in a RAID configuration, servers may have dual power supplies and hot-swappable CPUs, and datacenters may have batteries and diesel generators for backup power. When one component dies, the redundant component can take its place while the broken com‐ ponent is replaced. This approach cannot completely prevent hardware problems from causing failures, but it is well understood and can often keep a machine running uninterrupted for years. Reliability | 7 Until recently, redundancy of hardware components was sufficient for most applica‐ tions, since it makes total failure of a single machine fairly rare. As long as you can restore a backup onto a new machine fairly quickly, the downtime in case of failure is not catastrophic in most applications. Thus, multi-machine redundancy was only required by a small number of applications for which high availability was absolutely essential. However, as data volumes and applications’ computing demands have increased, more applications have begun using larger numbers of machines, which proportion‐ ally increases the rate of hardware faults. Moreover, in some cloud platforms such as Amazon Web Services (AWS) it is fairly common for virtual machine instances to become unavailable without warning , as the platforms are designed to prioritize flexibility and elasticityi over single-machine reliability. Hence there is a move toward systems that can tolerate the loss of entire machines, by using software fault-tolerance techniques in preference or in addition to hardware redundancy. Such systems also have operational advantages: a single-server system requires planned downtime if you need to reboot the machine (to apply operating system security patches, for example), whereas a system that can tolerate machine failure can be patched one node at a time, without downtime of the entire system (a rolling upgrade; see Chapter 4). Software Errors We usually think of hardware faults as being random and independent from each other: one machine’s disk failing does not imply that another machine’s disk is going to fail. There may be weak correlations (for example due to a common cause, such as the temperature in the server rack), but otherwise it is unlikely that a large number of hardware components will fail at the same time. Another class of fault is a systematic error within the system. Such faults are harder to anticipate, and because they are correlated across nodes, they tend to cause many more system failures than uncorrelated hardware faults. Examples include: A software bug that causes every instance of an application server to crash when given a particular bad input. For example, consider the leap second on June 30, 2012, that caused many applications to hang simultaneously due to a bug in the Linux kernel. A runaway process that uses up some shared resource—CPU time, memory, disk space, or network bandwidth. i. Defined in “Approaches for Coping with Load” on page 17. 8 | Chapter 1: Reliable, Scalable, and Maintainable Applications  A service that the system depends on that slows down, becomes unresponsive, or starts returning corrupted responses. Cascading failures, where a small fault in one component triggers a fault in another component, which in turn triggers further faults. The bugs that cause these kinds of software faults often lie dormant for a long time until they are triggered by an unusual set of circumstances. In those circumstances, it is revealed that the software is making some kind of assumption about its environ‐ ment—and while that assumption is usually true, it eventually stops being true for some reason. There is no quick solution to the problem of systematic faults in software. Lots of small things can help: carefully thinking about assumptions and interactions in the system; thorough testing; process isolation; allowing processes to crash and restart; measuring, monitoring, and analyzing system behavior in production. If a system is expected to provide some guarantee (for example, in a message queue, that the num‐ ber of incoming messages equals the number of outgoing messages), it can constantly check itself while it is running and raise an alert if a discrepancy is found. Human Errors Humans design and build software systems, and the operators who keep the systems running are also human. Even when they have the best intentions, humans are known to be unreliable. For example, one study of large internet services found that configuration errors by operators were the leading cause of outages, whereas hard‐ ware faults (servers or network) played a role in only 10–25% of outages. How do we make our systems reliable, in spite of unreliable humans? The best sys‐ tems combine several approaches: Design systems in a way that minimizes opportunities for error. For example, well-designed abstractions, APIs, and admin interfaces make it easy to do “the right thing” and discourage “the wrong thing.” However, if the interfaces are too restrictive people will work around them, negating their benefit, so this is a tricky balance to get right. Decouple the places where people make the most mistakes from the places where they can cause failures. In particular, provide fully featured non-production sandbox environments where people can explore and experiment safely, using real data, without affecting real users. Test thoroughly at all levels, from unit tests to whole-system integration tests and manual tests. Automated testing is widely used, well understood, and espe‐ cially valuable for covering corner cases that rarely arise in normal operation. Reliability | 9  Allow quick and easy recovery from human errors, to minimize the impact in the case of a failure. For example, make it fast to roll back configuration changes, roll out new code gradually (so that any unexpected bugs affect only a small subset of users), and provide tools to recompute data (in case it turns out that the old com‐ putation was incorrect). Set up detailed and clear monitoring, such as performance metrics and error rates. In other engineering disciplines this is referred to as telemetry. (Once a rocket has left the ground, telemetry is essential for tracking what is happening, and for understanding failures.) Monitoring can show us early warning sig‐ nals and allow us to check whether any assumptions or constraints are being vio‐ lated. When a problem occurs, metrics can be invaluable in diagnosing the issue. Implement good management practices and training—a complex and important aspect, and beyond the scope of this book. How Important Is Reliability? Reliability is not just for nuclear power stations and air traffic control software— more mundane applications are also expected to work reliably. Bugs in business applications cause lost productivity (and legal risks if figures are reported incor‐ rectly), and outages of ecommerce sites can have huge costs in terms of lost revenue and damage to reputation. Even in “noncritical” applications we have a responsibility to our users. Consider a parent who stores all their pictures and videos of their children in your photo appli‐ cation. How would they feel if that database was suddenly corrupted? Would they know how to restore it from a backup? There are situations in which we may choose to sacrifice reliability in order to reduce development cost (e.g., when developing a prototype product for an unproven mar‐ ket) or operational cost (e.g., for a service with a very narrow profit margin)—but we should be very conscious of when we are cutting corners. Scalability Even if a system is working reliably today, that doesn’t mean it will necessarily work reliably in the future. One common reason for degradation is increased load: perhaps the system has grown from 10,000 concurrent users to 100,000 concurrent users, or from 1 million to 10 million. Perhaps it is processing much larger volumes of data than it did before. Scalability is the term we use to describe a system’s ability to cope with increased load. Note, however, that it is not a one-dimensional label that we can attach to a sys‐ tem: it is meaningless to say “X is scalable” or “Y doesn’t scale.” Rather, discussing 10 | Chapter 1: Reliable, Scalable, and Maintainable Applications scalability means considering questions like “If the system grows in a particular way, what are our options for coping with the growth?” and “How can we add computing resources to handle the additional load?” Describing Load First, we need to succinctly describe the current load on the system; only then can we discuss growth questions (what happens if our load doubles?). Load can be described with a few numbers which we call load parameters. The best choice of parameters depends on the architecture of your system: it may be requests per second to a web server, the ratio of reads to writes in a database, the number of simultaneously active users in a chat room, the hit rate on a cache, or something else. Perhaps the average case is what matters for you, or perhaps your bottleneck is dominated by a small number of extreme cases. To make this idea more concrete, let’s consider Twitter as an example, using data published in November 2012. Two of Twitter’s main operations are: Post tweet A user can publish a new message to their followers (4.6k requests/sec on aver‐ age, over 12k requests/sec at peak). Home timeline A user can view tweets posted by the people they follow (300k requests/sec). Simply handling 12,000 writes per second (the peak rate for posting tweets) would be fairly easy. However, Twitter’s scaling challenge is not primarily due to tweet volume, but due to fan-outii—each user follows many people, and each user is followed by many people. There are broadly two ways of implementing these two operations: 1. Posting a tweet simply inserts the new tweet into a global collection of tweets. When a user requests their home timeline, look up all the people they follow, find all the tweets for each of those users, and merge them (sorted by time). In a relational database like in Figure 1-2, you could write a query such as: SELECT tweets.*, users.* FROM tweets JOIN users ON tweets.sender_id = users.id JOIN follows ON follows.followee_id = users.id WHERE follows.follower_id = current_user ii. A term borrowed from electronic engineering, where it describes the number of logic gate inputs that are attached to another gate’s output. The output needs to supply enough current to drive all the attached inputs. In transaction processing systems, we use it to describe the number of requests to other services that we need to make in order to serve one incoming request. Scalability | 11  2. Maintain a cache for each user’s home timeline—like a mailbox of tweets for each recipient user (see Figure 1-3). When a user posts a tweet, look up all the people who follow that user, and insert the new tweet into each of their home timeline caches. The request to read the home timeline is then cheap, because its result has been computed ahead of time. Figure 1-2. Simple relational schema for implementing a Twitter home timeline. Figure 1-3. Twitter’s data pipeline for delivering tweets to followers, with load parame‐ ters as of November 2012. The first version of Twitter used approach 1, but the systems struggled to keep up with the load of home timeline queries, so the company switched to approach 2. This works better because the average rate of published tweets is almost two orders of magnitude lower than the rate of home timeline reads, and so in this case it’s prefera‐ ble to do more work at write time and less at read time. However, the downside of approach 2 is that posting a tweet now requires a lot of extra work. On average, a tweet is delivered to about 75 followers, so 4.6k tweets per second become 345k writes per second to the home timeline caches. But this average hides the fact that the number of followers per user varies wildly, and some users 12 | Chapter 1: Reliable, Scalable, and Maintainable Applications have over 30 million followers. This means that a single tweet may result in over 30 million writes to home timelines! Doing this in a timely manner—Twitter tries to deliver tweets to followers within five seconds—is a significant challenge. In the example of Twitter, the distribution of followers per user (maybe weighted by how often those users tweet) is a key load parameter for discussing scalability, since it determines the fan-out load. Your application may have very different characteristics, but you can apply similar principles to reasoning about its load. The final twist of the Twitter anecdote: now that approach 2 is robustly implemented, Twitter is moving to a hybrid of both approaches. Most users’ tweets continue to be fanned out to home timelines at the time when they are posted, but a small number of users with a very large number of followers (i.e., celebrities) are excepted from this fan-out. Tweets from any celebrities that a user may follow are fetched separately and merged with that user’s home timeline when it is read, like in approach 1. This hybrid approach is able to deliver consistently good performance. We will revisit this exam‐ ple in Chapter 12 after we have covered some more technical ground. Describing Performance Once you have described the load on your system, you can investigate what happens when the load increases. You can look at it in two ways: When you increase a load parameter and keep the system resources (CPU, mem‐ ory, network bandwidth, etc.) unchanged, how is the performance of your system affected? When you increase a load parameter, how much do you need to increase the resources if you want to keep performance unchanged? Both questions require performance numbers, so let’s look briefly at describing the performance of a system. In a batch processing system such as Hadoop, we usually care about throughput—the number of records we can process per second, or the total time it takes to run a job on a dataset of a certain size.iii In online systems, what’s usually more important is the service’s response time—that is, the time between a client sending a request and receiving a response. iii. In an ideal world, the running time of a batch job is the size of the dataset divided by the throughput. In practice, the running time is often longer, due to skew (data not being spread evenly across worker processes) and needing to wait for the slowest task to complete. Scalability | 13  Latency and response time Latency and response time are often used synonymously, but they are not the same. The response time is what the client sees: besides the actual time to process the request (the service time), it includes network delays and queueing delays. Latency is the duration that a request is waiting to be handled—during which it is latent, await‐ ing service. Even if you only make the same request over and over again, you’ll get a slightly dif‐ ferent response time on every try. In practice, in a system handling a variety of requests, the response time can vary a lot. We therefore need to think of response time not as a single number, but as a distribution of values that you can measure. In Figure 1-4, each gray bar represents a request to a service, and its height shows how long that request took. Most requests are reasonably fast, but there are occa‐ sional outliers that take much longer. Perhaps the slow requests are intrinsically more expensive, e.g., because they process more data. But even in a scenario where you’d think all requests should take the same time, you get variation: random additional latency could be introduced by a context switch to a background process, the loss of a network packet and TCP retransmission, a garbage collection pause, a page fault forcing a read from disk, mechanical vibrations in the server rack , or many other causes. Figure 1-4. Illustrating mean and percentiles: response times for a sample of 100 requests to a service. It’s common to see the average response time of a service reported. (Strictly speaking, the term “average” doesn’t refer to any particular formula, but in practice it is usually understood as the arithmetic mean: given n values, add up all the values, and divide by n.) However, the mean is not a very good metric if you want to know your “typi‐ cal” response time, because it doesn’t tell you how many users actually experienced that delay. Usually it is better to use percentiles. If you take your list of response times and sort it from fastest to slowest, then the median is the halfway point: for example, if your 14 | Chapter 1: Reliable, Scalable, and Maintainable Applications median response time is 200 ms, that means half your requests return in less than 200 ms, and half your requests take longer than that. This makes the median a good metric if you want to know how long users typically have to wait: half of user requests are served in less than the median response time, and the other half take longer than the median. The median is also known as the 50th percentile, and sometimes abbreviated as p50. Note that the median refers to a single request; if the user makes several requests (over the course of a session, or because several resources are included in a single page), the probability that at least one of them is slower than the median is much greater than 50%. In order to figure out how bad your outliers are, you can look at higher percentiles: the 95th, 99th, and 99.9th percentiles are common (abbreviated p95, p99, and p999). They are the response time thresholds at which 95%, 99%, or 99.9% of requests are faster than that particular threshold. For example, if the 95th percentile response time is 1.5 seconds, that means 95 out of 100 requests take less than 1.5 seconds, and 5 out of 100 requests take 1.5 seconds or more. This is illustrated in Figure 1-4. High percentiles of response times, also known as tail latencies, are important because they directly affect users’ experience of the service. For example, Amazon describes response time requirements for internal services in terms of the 99.9th per‐ centile, even though it only affects 1 in 1,000 requests. This is because the customers with the slowest requests are often those who have the most data on their accounts because they have made many purchases—that is, they’re the most valuable custom‐ ers. It’s important to keep those customers happy by ensuring the website is fast for them: Amazon has also observed that a 100 ms increase in response time reduces sales by 1% , and others report that a 1-second slowdown reduces a customer sat‐ isfaction metric by 16% [21, 22]. On the other hand, optimizing the 99.99th percentile (the slowest 1 in 10,000 requests) was deemed too expensive and to not yield enough benefit for Amazon’s purposes. Reducing response times at very high percentiles is difficult because they are easily affected by random events outside of your control, and the benefits are diminishing. For example, percentiles are often used in service level objectives (SLOs) and service level agreements (SLAs), contracts that define the expected performance and availa‐ bility of a service. An SLA may state that the service is considered to be up if it has a median response time of less than 200 ms and a 99th percentile under 1 s (if the response time is longer, it might as well be down), and the service may be required to be up at least 99.9% of the time. These metrics set expectations for clients of the ser‐ vice and allow customers to demand a refund if the SLA is not met. Queueing delays often account for a large part of the response time at high percen‐ tiles. As a server can only process a small number of things in parallel (limited, for Scalability | 15 example, by its number of CPU cores), it only takes a small number of slow requests to hold up the processing of subsequent requests—an effect sometimes known as head-of-line blocking. Even if those subsequent requests are fast to process on the server, the client will see a slow overall response time due to the time waiting for the prior request to complete. Due to this effect, it is important to measure response times on the client side. When generating load artificially in order to test the scalability of a system, the load- generating client needs to keep sending requests independently of the response time. If the client waits for the previous request to complete before sending the next one, that behavior has the effect of artificially keeping the queues shorter in the test than they would be in reality, which skews the measurements. Percentiles in Practice High percentiles become especially important in backend services that are called mul‐ tiple times as part of serving a single end-user request. Even if you make the calls in parallel, the end-user request still needs to wait for the slowest of the parallel calls to complete. It takes just one slow call to make the entire end-user request slow, as illus‐ trated in Figure 1-5. Even if only a small percentage of backend calls are slow, the chance of getting a slow call increases if an end-user request requires multiple back‐ end calls, and so a higher proportion of end-user requests end up being slow (an effect known as tail latency amplification ). If you want to add response time percentiles to the monitoring dashboards for your services, you need to efficiently calculate them on an ongoing basis. For example, you may want to keep a rolling window of response times of requests in the last 10 minutes. Every minute, you calculate the median and various percentiles over the val‐ ues in that window and plot those metrics on a graph. The naïve implementation is to keep a list of response times for all requests within the time window and to sort that list every minute. If that is too inefficient for you, there are algorithms that can calculate a good approximation of percentiles at minimal CPU and memory cost, such as forward decay , t-digest , or HdrHistogram. Beware that averaging percentiles, e.g., to reduce the time resolution or to com‐ bine data from several machines, is mathematically meaningless—the right way of aggregating response time data is to add the histograms. 16 | Chapter 1: Reliable, Scalable, and Maintainable Applications Figure 1-5. When several backend calls are needed to serve a request, it takes just a sin‐ gle slow backend request to slow down the entire end-user request. Approaches for Coping with Load Now that we have discussed the parameters for describing load and metrics for meas‐ uring performance, we can start discussing scalability in earnest: how do we maintain good performance even when our load parameters increase by some amount? An architecture that is appropriate for one level of load is unlikely to cope with 10 times that load. If you are working on a fast-growing service, it is therefore likely that you will need to rethink your architecture on every order of magnitude load increase —or perhaps even more often than that. People often talk of a dichotomy between scaling up (vertical scaling, moving to a more powerful machine) and scaling out (horizontal scaling, distributing the load across multiple smaller machines). Distributing load across multiple machines is also known as a shared-nothing architecture. A system that can run on a single machine is often simpler, but high-end machines can become very expensive, so very intensive workloads often can’t avoid scaling out. In reality, good architectures usually involve a pragmatic mixture of approaches: for example, using several fairly powerful machines can still be simpler and cheaper than a large number of small virtual machines. Some systems are elastic, meaning that they can automatically add computing resour‐ ces when they detect a load increase, whereas other systems are scaled manually (a human analyzes the capacity and decides to add more machines to the system). An elastic system can be useful if load is highly unpredictable, but manually scaled sys‐ tems are simpler and may have fewer operational surprises (see “Rebalancing Parti‐ tions” on page 209). Scalability | 17 While distributing stateless services across multiple machines is fairly straightfor‐ ward, taking stateful data systems from a single node to a distributed setup can intro‐ duce a lot of additional complexity. For this reason, common wisdom until recently was to keep your database on a single node (scale up) until scaling cost or high- availability requirements forced you to make it distributed. As the tools and abstractions for distributed systems get better, this common wisdom may change, at least for some kinds of applications. It is conceivable that distributed data systems will become the default in the future, even for use cases that don’t han‐ dle large volumes of data or traffic. Over the course of the rest of this book we will cover many kinds of distributed data systems, and discuss how they fare not just in terms of scalability, but also ease of use and maintainability. The architecture of systems that operate at large scale is usually highly specific to the application—there is no such thing as a generic, one-size-fits-all scalable architecture (informally known as magic scaling sauce). The problem may be the volume of reads, the volume of writes, the volume of data to store, the complexity of the data, the response time requirements, the access patterns, or (usually) some mixture of all of these plus many more issues. For example, a system that is designed to handle 100,000 requests per second, each 1 kB in size, looks very different from a system that is designed for 3 requests per minute, each 2 GB in size—even though the two systems have the same data through‐ put. An architecture that scales well for a particular application is built around assump‐ tions of which operations will be common and which will be rare—the load parame‐ ters. If those assumptions turn out to be wrong, the engineering effort for scaling is at best wasted, and at worst counterproductive. In an early-stage startup or an unpro‐ ven product it’s usually more important to be able to iterate quickly on product fea‐ tures than it is to scale to some hypothetical future load. Even though they are specific to a particular application, scalable architectures are nevertheless usually built from general-purpose building blocks, arranged in familiar patterns. In this book we discuss those building blocks and patterns. Maintainability It is well known that the majority of the cost of software is not in its initial develop‐ ment, but in its ongoing maintenance—fixing bugs, keeping its systems operational, investigating failures, adapting it to new platforms, modifying it for new use cases, repaying technical debt, and adding new features. Yet, unfortunately, many people working on software systems dislike maintenance of so-called legacy systems—perhaps it involves fixing other people’s mistakes, or work‐ 18 | Chapter 1: Reliable, Scalable, and Maintainable Applications ing with platforms that are now outdated, or systems that were forced to do things they were never intended for. Every legacy system is unpleasant in its own way, and so it is difficult to give general recommendations for dealing with them. However, we can and should design software in such a way that it will hopefully min‐ imize pain during maintenance, and thus avoid creating legacy software ourselves. To this end, we will pay particular attention to three design principles for software systems: Operability Make it easy for operations teams to keep the system running smoothly. Simplicity Make it easy for new engineers to understand the system, by removing as much complexity as possible from the system. (Note this is not the same as simplicity of the user interface.) Evolvability Make it easy for engineers to make changes to the system in the future, adapting it for unanticipated use cases as requirements change. Also known as extensibil‐ ity, modifiability, or plasticity. As previously with reliability and scalability, there are no easy solutions for achieving these goals. Rather, we will try to think about systems with operability, simplicity, and evolvability in mind. Operability: Making Life Easy for Operations It has been suggested that “good operations can often work around the limitations of bad (or incomplete) software, but good software cannot run reliably with bad opera‐ tions”. While some aspects of operations can and should be automated, it is still up to humans to set up that automation in the first place and to make sure it’s work‐ ing correctly. Operations teams are vital to keeping a software system running smoothly. A good operations team typically is responsible for the following, and more : Monitoring the health of the system and quickly restoring service if it goes into a bad state Tracking down the cause of problems, such as system failures or degraded per‐ formance Keeping software and platforms up to date, including security patches Keeping tabs on how different systems affect each other, so that a problematic change can be avoided before it causes damage Maintainability | 19  Anticipating future problems and solving them before they occur (e.g., capacity planning) Establishing good practices and tools for deployment, configuration manage‐ ment, and more Performing complex maintenance tasks, such as moving an application from one platform to another Maintaining the security of the system as configuration changes are made Defining processes that make operations predictable and help keep the produc‐ tion environment stable Preserving the organization’s knowledge about the system, even as individual people come and go Good operability means making routine tasks easy, allowing the operations team to focus their efforts on high-value activities. Data systems can do various things to make routine tasks easy, including: Providing visibility into the runtime behavior and internals of the system, with good monitoring Providing good support for automation and integration with standard tools Avoiding dependency on individual machines (allowing machines to be taken down for maintenance while the system as a whole continues running uninter‐ rupted) Providing good documentation and an easy-to-understand operational model (“If I do X, Y will happen”) Providing good default behavior, but also giving administrators the freedom to override defaults when needed Self-healing where appropriate, but also giving administrators manual control over the system state when needed Exhibiting predictable behavior, minimizing surprises Simplicity: Managing Complexity Small software projects can have delightfully simple and expressive code, but as projects get larger, they often become very complex and difficult to understand. This complexity slows down everyone who needs to work on the system, further increas‐ ing the cost of maintenance. A software project mired in complexity is sometimes described as a big ball of mud. 20 | Chapter 1: Reliable, Scalable, and Maintainable Applications There are various possible symptoms of complexity: explosion of the state space, tight coupling of modules, tangled dependencies, inconsistent naming and terminology, hacks aimed at solving performance problems, special-casing to work around issues elsewhere, and many more. Much has been said on this topic already [31, 32, 33]. When complexity makes maintenance hard, budgets and schedules are often over‐ run. In complex software, there is also a greater risk of introducing bugs when mak‐ ing a change: when the system is harder for developers to understand and reason about, hidden assumptions, unintended consequences, and unexpected interactions are more easily overlooked. Conversely, reducing complexity greatly improves the maintainability of software, and thus simplicity should be a key goal for the systems we build. Making a system simpler does not necessarily mean reducing its functionality; it can also mean removing accidental complexity. Moseley and Marks define complex‐ ity as accidental if it is not inherent in the problem that the software solves (as seen by the users) but arises only from the implementation. One of the best tools we have for removing accidental complexity is abstraction. A good abstraction can hide a great deal of implementation detail behind a clean, simple-to-understand façade. A good abstraction can also be used for a wide range of different applications. Not only is this reuse more efficient than reimplementing a similar thing multiple times, but it also leads to higher-quality software, as quality improvements in the abstracted component benefit all applications that use it. For example, high-level programming languages are abstractions that hide machine code, CPU registers, and syscalls. SQL is an abstraction that hides complex on-disk and in-memory data structures, concurrent requests from other clients, and inconsis‐ tencies after crashes. Of course, when programming in a high-level language, we are still using machine code; we are just not using it directly, because the programming language abstraction saves us from having to think about it. However, finding good abstractions is very hard. In the field of distributed systems, although there are many good algorithms, it is much less clear how we should be packaging them into abstractions that help us keep the complexity of the system at a manageable level. Throughout this book, we will keep our eyes open for good abstractions that allow us to extract parts of a large system into well-defined, reusable components. Evolvability: Making Change Easy It’s extremely unlikely that your system’s requirements will remain unchanged for‐ ever. They are much more likely to be in constant flux: you learn new facts, previ‐ ously unanticipated use cases emerge, business priorities change, users request new Maintainability | 21 features, new platforms replace old platforms, legal or regulatory requirements change, growth of the system forces architectural changes, etc. In terms of organizational processes, Agile working patterns provide a framework for adapting to change. The Agile community has also developed technical tools and pat‐ terns that are helpful when developing software in a frequently changing environ‐ ment, such as test-driven development (TDD) and refactoring. Most discussions of these Agile techniques focus on a fairly small, local scale (a cou‐ ple of source code files within the same application). In this book, we search for ways of increasing agility on the level of a larger data system, perhaps consisting of several different applications or services with different characteristics. For example, how would you “refactor” Twitter’s architecture for assembling home timelines (“Describ‐ ing Load” on page 11) from approach 1 to approach 2? The ease with which you can modify a data system, and adapt it to changing require‐ ments, is closely linked to its simplicity and its abstractions: simple and easy-to- understand systems are usually easier to modify than complex ones. But since this is such an important idea, we will use a different word to refer to agility on a data sys‐ tem level: evolvability. Summary In this chapter, we have explored some fundamental ways of thinking about data- intensive applications. These principles will guide us through the rest of the book, where we dive into deep technical detail. An application has to meet various requirements in order to be useful. There are functional requirements (what it should do, such as allowing data to be stored, retrieved, searched, and processed in various ways), and some nonfunctional require‐ ments (general properties like security, reliability, compliance, scalability, compatibil‐ ity, and maintainability). In this chapter we discussed reliability, scalability, and maintainability in detail. Reliability means making systems work correctly, even when faults occur. Faults can be in hardware (typically random and uncorrelated), software (bugs are typically sys‐ tematic and hard to deal with), and humans (who inevitably make mistakes from time to time). Fault-tolerance techniques can hide certain types of faults from the end user. Scalability means having strategies for keeping performance good, even when load increases. In order to discuss scalability, we first need ways of describing load and performance quantitatively. We briefly looked at Twitter’s home timelines as an example of describing load, and response time percentiles as a way of measuring per‐ 22 | Chapter 1: Reliable, Scalable, and Maintainable Applications formance. In a scalable system, you can add processing capacity in order to remain reliable under high load. Maintainability has many facets, but in essence it’s about making life better for the engineering and operations teams who need to work with the system. Good abstrac‐ tions can help reduce complexity and make the system easier to modify and adapt for new use cases. Good operability means having good visibility into the system’s health, and having effective ways of managing it. There is unfortunately no easy fix for making applications reliable, scalable, or main‐ tainable. However, there are certain patterns and techniques that keep reappearing in different kinds of applications. In the next few chapters we will take a look at some examples of data systems and analyze how they work toward those goals. Later in the book, in Part III, we will look at patterns for systems that consist of sev‐ eral components working together, such as the one in Figure 1-1. References Michael Stonebraker and Uğur Çetintemel: “‘One Size Fits All’: An Idea Whose Time Has Come and Gone,” at 21st International Conference on Data Engineering (ICDE), April 2005. Walter L. Heimerdinger and Charles B. Weinstock: “A Conceptual Framework for System Fault Tolerance,” Technical Report CMU/SEI-92-TR-033, Software Engi‐ neering Institute, Carnegie Mellon University, October 1992. Ding Yuan, Yu Luo, Xin Zhuang, et al.: “Simple Testing Can Prevent Most Criti‐ cal Failures: An Analysis of Production Failures in Distributed Data-Intensive Sys‐ tems,” at 11th USENIX Symposium on Operating Systems Design and Implementation (OSDI), October 2014. Yury Izrailevsky and Ariel Tseitlin: “The Netflix Simian Army,” techblog.net‐ flix.com, July 19, 2011. Daniel Ford, François Labelle, Florentina I. Popovici, et al.: “Availability in Glob‐ ally Distributed Storage Systems,” at 9th USENIX Symposium on Operating Systems Design and Implementation (OSDI), October 2010. Brian Beach: “Hard Drive Reliability Update – Sep 2014,” backblaze.com, Septem‐ ber 23, 2014.