If you are building an application, you will be driven by a list of requirements. At the top of your list is most likely the functionality that the application must offer: what screens and what buttons you need, and what each operation is supposed to do in order to fulfill the purpose of your software. These are your *functional requirements*.
In addition, you probably also have some *nonfunctional requirements*: for example, the app should be fast, reliable, secure, legally compliant, and easy to maintain. These requirements might not be explicitly written down, because they may seem somewhat obvious, but they are just as important as the app’s functionality: an app that is unbearably slow or unreliable might as well not exist.
Not all nonfunctional requirements fall within the scope of this book, but several do. In this chapter we will introduce several technical concepts that will help you articulate the nonfunctional requirements for your own systems:
- How to define and measure the *performance* of a system (see [“Describing Performance”](ch02.html#sec_introduction_percentiles));
- What it means for a service to be *reliable*—namely, continuing to work correctly, even when things go wrong (see [“Reliability and Fault Tolerance”](ch02.html#sec_introduction_reliability));
- Allowing a system to be *scalable* by having efficient ways of adding computing capacity as the load on the system grows (see [“Scalability”](ch02.html#sec_introduction_scalability)); and
- Making it easier to maintain a system in the long term (see [“Maintainability”](ch02.html#sec_introduction_maintainability)).
The terminology introduced in this chapter will also be useful in the following chapters, when we go into the details of how data-intensive systems are implemented. However, abstract definitions can be quite dry; to make the ideas more concrete, we will start this chapter with a case study of how a social networking service might work, which will provide practical examples of performance and scalability.
Imagine you are given the task of implementing a social network in the style of X (formerly Twitter), in which users can post messages and follow other users. This will be a huge simplification of how such a service actually works [[1](ch02.html#Cvet2016), [2](ch02.html#Krikorian2012_ch2), [3](ch02.html#Twitter2023)], but it will help illustrate some of the issues that arise in large-scale systems.
Let’s assume that users make 500 million posts per day, or 5,700 posts per second on average. Occasionally, the rate can spike as high as 150,000 posts/second [[4](ch02.html#Krikorian2013)]. Let’s also assume that the average user follows 200 people and has 200 followers (although there is a very wide range: most people have only a handful of followers, and a few celebrities such as Barack Obama have over 100 million followers).
Imagine we keep all of the data in a relational database as shown in [Figure 2-1](ch02.html#fig_twitter_relational). We have one table for users, one table for posts, and one table for follow relationships.
Let’s say the main read operation that our social network must support is the *home timeline*, which displays recent posts by people you are following (for simplicity we will ignore ads, suggested posts from people you are not following, and other extensions). We could write the following SQL query to get the home timeline for a particular user:
To execute this query, the database will use the `follows` table to find everybody who `current_user` is following, look up recent posts by those users, and sort them by timestamp to get the most recent 1,000 posts by any of the followed users.
Posts are supposed to be timely, so let’s assume that after somebody makes a post, we want their followers to be able to see it within 5 seconds. One way of doing that would be for the user’s client to repeat the query above every 5 seconds while the user is online (this is known as *polling*). If we assume that 10 million users are online and logged in at the same time, that would mean running the query 2 million times per second. Even if you increase the polling interval, this is a lot.
Moreover, the query above is quite expensive: if you are following 200 people, it needs to fetch a list of recent posts by each of those 200 people, and merge those lists. 2 million timeline queries per second then means that the database needs to look up the recent posts from some sender 400 million times per second—a huge number. And that is the average case. Some users follow tens of thousands of accounts; for them, this query is very expensive to execute, and difficult to make fast.
How can we do better? Firstly, instead of polling, it would be better if the server actively pushed new posts to any followers who are currently online. Secondly, we should precompute the results of the query above so that a user’s request for their home timeline can be served from a cache.
Imagine that for each user we store a data structure containing their home timeline, i.e., the recent posts by people they are following. Every time a user makes a post, we look up all of their followers, and insert that post into the home timeline of each follower—like delivering a message to a mailbox. Now when a user logs in, we can simply give them this home timeline that we precomputed. Moreover, to receive a notification about any new posts on their timeline, the user’s client simply needs to subscribe to the stream of posts being added to their home timeline.
The downside of this approach is that we now need to do more work every time a user makes a post, because the home timelines are derived data that needs to be updated. The process is illustrated in [Figure 2-2](ch02.html#fig_twitter_timelines). When one initial request results in several downstream requests being carried out, we use the term *fan-out* to describe the factor by which the number of requests increases.
At a rate of 5,700 posts posted per second, if the average post reaches 200 followers (i.e., a fan-out factor of 200), we will need to do just over 1 million home timeline writes per second. This is a lot, but it’s still a significant saving compared to the 400 million per-sender post lookups per second that we would otherwise have to do.
If the rate of posts spikes due to some special event, we don’t have to do the timeline deliveries immediately—we can enqueue them and accept that it will temporarily take a bit longer for posts to show up in followers’ timelines. Even during such load spikes, timelines remain fast to load, since we simply serve them from a cache.
This process of precomputing and updating the results of a query is called *materialization*, and the timeline cache is an example of a *materialized view* (a concept we will discuss further in [Link to Come]). The downside of materialization is that every time a celebrity makes a post, we now have to do a large amount of work to insert that post into the home timelines of each of their millions of followers.
One way of solving this problem is to handle celebrity posts separately from everyone else’s posts: we can save ourselves the effort of adding them to millions of timelines by storing the celebrity posts separately and merging them with the materialized timeline when it is read. Despite such optimizations, handling celebrities on a social network can require a lot of infrastructure [[5](ch02.html#Axon2010_ch2)].
The number of requests per second, or the data volume per second, that the system is processing. For a given a particular allocation of hardware resources, there is a *maximum throughput* that can be handled. The unit of measurement is “somethings per second”.
In the social network case study, “posts per second” and “timeline writes per second” are throughput metrics, whereas the “time it takes to load the home timeline” or the “time until a post is delivered to followers” are response time metrics.
There is often a connection between throughput and response time; an example of such a relationship for an online service is sketched in [Figure 2-3](ch02.html#fig_throughput). The service has a low response time when request throughput is low, but response time increases as load increases. This is because of *queueing*: when a request arrives on a highly loaded system, it’s likely that the CPU is already in the process of handling an earlier request, and therefore the incoming request needs to wait until the earlier request has been completed. As throughput approaches the maximum that the hardware can handle, queueing delays increase sharply.
If a system is close to overload, with throughput pushed close to the limit, it can sometimes enter a vicious cycle where it becomes less efficient and hence even more overloaded. For example, if there is a long queue of requests waiting to be handled, response times may increase so much that clients time out and resend their request. This causes the rate of requests to increase even further, making the problem worse—a *retry storm*. Even when the load is reduced again, such a system may remain in an overloaded state until it is rebooted or otherwise reset. This phenomenon is called a *metastable failure*, and it can cause serious outages in production systems [[6](ch02.html#Bronson2021), [7](ch02.html#Brooker2021)].
To avoid retries overloading a service, you can increase and randomize the time between successive retries on the client side (*exponential backoff* [[8](ch02.html#Brooker2015), [9](ch02.html#Brooker2022backoff)]), and temporarily stop sending requests to a service that has returned errors or timed out recently (using a *circuit breaker* [[10](ch02.html#Nygard2018)] or *token bucket* algorithm [[11](ch02.html#Brooker2022retries)]). The server can also detect when it is approaching overload and start proactively rejecting requests (*load shedding* [[12](ch02.html#YanacekLoadShedding)]), and send back responses asking clients to slow down (*backpressure* [[1](ch02.html#Cvet2016), [13](ch02.html#Sackman2016_ch2)]). The choice of queueing and load-balancing algorithms can also make a difference [[14](ch02.html#Kopytkov2018)].
In terms of performance metrics, the response time is usually what users care about the most, whereas the throughput determines the required computing resources (e.g., how many servers you need), and hence the cost of serving a particular workload. If throughput is likely to increase beyond what the current hardware can handle, the capacity needs to be expanded; a system is said to be *scalable* if its maximum throughput can be significantly increased by adding computing resources.
In this section we will focus primarily on response times, and we will return to throughput and scalability in [“Scalability”](ch02.html#sec_introduction_scalability).
“Latency” and “response time” are sometimes used interchangeably, but in this book we will use the terms in a specific way (illustrated in [Figure 2-4](ch02.html#fig_response_time)):
- The *response time* is what the client sees; it includes all delays incurred anywhere in the system.
- The *service time* is the duration for which the service is actively processing the user request.
- *Queueing delays* can occur at several points in the flow: for example, after a request is received, it might need to wait until a CPU is available before it can be processed; a response packet might need to be buffered before it is sent over the network if other tasks on the same machine are sending a lot of data via the outbound network interface.
- *Latency* is a catch-all term for time during which a request is not being actively processed, i.e., during which it is *latent*. In particular, *network latency* or *network delay* refers to the time that request and response spend traveling through the network.
排隊延遲通常是響應時間變化性的一個重要部分。由於伺服器同時只能處理少量事務（例如，受其 CPU 核心數量的限制），只需少數幾個慢請求就足以阻塞後續請求的處理——這種效應被稱為 *隊首阻塞*。即使那些後續請求的服務時間很快，客戶端也會因為等待先前請求完成而感覺到整體響應時間的緩慢。排隊延遲不屬於服務時間的一部分，因此在客戶端測量響應時間十分重要。
The response time can vary significantly from one request to the next, even if you keep making the same request over and over again. Many factors can add random delays: for example, 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 [[15](ch02.html#Gunawi2018)], or many other causes. We will discuss this topic in more detail in [Link to Come].
Queueing delays often account for a large part of the variability in response times. As a server can only process a small number of things in parallel (limited, for 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 known as *head-of-line blocking*. Even if those subsequent requests have fast service times, the client will see a slow overall response time due to the time waiting for the prior request to complete. The queueing delay is not part of the service time, and for this reason it is important to measure response times on the client side.
Because the response time varies from one request to the next, we need to think of it not as a single number, but as a *distribution* of values that you can measure. In [Figure 2-5](ch02.html#fig_lognormal), 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 occasional *outliers* that take much longer. Variation in network delay is also known as *jitter*.
It’s common to report the *average* response time of a service (technically, the *arithmetic mean*: that is, sum all the response times, and divide by the number of requests). However, the mean is not a very good metric if you want to know your “typical” 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 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. The median is also known as the *50th percentile*, and sometimes abbreviated as *p50*.
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 2-5](ch02.html#fig_lognormal).
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 percentile, 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 customers [[16](ch02.html#DeCandia2007_ch1)]. It’s important to keep those customers happy by ensuring the website is fast for them.
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.
It seems intuitively obvious that a fast service is better for users than a slow service [[17](ch02.html#Whitenton2020)]. However, it is surprisingly difficult to get hold of reliable data to quantify the effect that latency has on user behavior.
Some often-cited statistics are unreliable. In 2006 Google reported that a slowdown in search results from 400 ms to 900 ms was associated with a 20% drop in traffic and revenue [[18](ch02.html#Linden2006)]. However, another Google study from 2009 reported that a 400 ms increase in latency resulted in only 0.6% fewer searches per day [[19](ch02.html#Brutlag2009)], and in the same year Bing found that a two-second increase in load time reduced ad revenue by 4.3% [[20](ch02.html#Schurman2009)]. Newer data from these companies appears not to be publicly available.
A more recent Akamai study [[21](ch02.html#Akamai2017)] claims that a 100 ms increase in response time reduced the conversion rate of e-commerce sites by up to 7%; however, on closer inspection, the same study reveals that very *fast* page load times are also correlated with lower conversion rates! This seemingly paradoxical result is explained by the fact that the pages that load fastest are often those that have no useful content (e.g., 404 error pages). However, since the study makes no effort to separate the effects of page content from the effects of load time, its results are probably not meaningful.
A study by Yahoo [[22](ch02.html#Bai2017)] compares click-through rates on fast-loading versus slow-loading search results, controlling for quality of search results. It finds 20–30% more clicks on fast searches when the difference between fast and slow responses is 1.25 seconds or more.
High percentiles are especially important in backend services that are called multiple 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 illustrated in [Figure 2-6](ch02.html#fig_tail_amplification). 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 backend calls, and so a higher proportion of end-user requests end up being slow (an effect known as *tail latency amplification* [[23](ch02.html#Dean2013)]).
Percentiles are often used in *service level objectives* (SLOs) and *service level agreements* (SLAs) as ways of defining the expected performance and availability of a service [[24](ch02.html#Hidalgo2020)]. For example, an SLO may set a target for a service to have a median response time of less than 200 ms and a 99th percentile under 1 s, and a target that at least 99.9% of valid requests result in non-error responses. An SLA is a contract that specifies what happens if the SLO is not met (for example, customers may be entitled to a refund). That is the basic idea, at least; in practice, defining good availability metrics for SLOs and SLAs is not straightforward [[25](ch02.html#Mogul2019), [26](ch02.html#Hauer2020)].
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 values in that window and plot those metrics on a graph.
The simplest 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. Open source percentile estimation libraries include HdrHistogram, t-digest [[27](ch02.html#Dunning2021), [28](ch02.html#Kohn2021)], OpenHistogram [[29](ch02.html#Hartmann2020)], and DDSketch [[30](ch02.html#Masson2019)].
Beware that averaging percentiles, e.g., to reduce the time resolution or to combine data from several machines, is mathematically meaningless—the right way of aggregating response time data is to add the histograms [[31](ch02.html#Schwartz2015)].
A fault is when a particular *part* of a system stops working correctly: for example, if a single hard drive malfunctions, or a single machine crashes, or an external service (that the system depends on) has an outage.
A failure is when the system *as a whole* stops providing the required service to the user; in other words, when it does not meet the service level objective (SLO).
The distinction between fault and failure can be confusing because they are the same thing, just at different levels. For example, if a hard drive stops working, we say that the hard drive has failed: if the system consists only of that one hard drive, it has stopped providing the required service. However, if the system you’re talking about contains many hard drives, then the failure of a single hard drive is only a fault from the point of view of the bigger system, and the bigger system might be able to tolerate that fault by having a copy of the data on another hard drive.
We call a system *fault-tolerant* if it continues providing the required service to the user in spite of certain faults occurring. If a system cannot tolerate a certain part becoming faulty, we call that part a *single point of failure* (SPOF), because a fault in that part escalates to cause the failure of the whole system.
For example, in the social network case study, a fault that might happen is that during the fan-out process, a machine involved in updating the materialized timelines crashes or become unavailable. To make this process fault-tolerant, we would need to ensure that another machine can take over this task without missing any posts that should have been delivered, and without duplicating any posts. (This idea is known as *exactly-once semantics*, and we will examine it in detail in [Link to Come].)
Fault tolerance is always limited to a certain number of certain types of faults. For example, a system might be able to tolerate a maximum of two hard drives failing at the same time, or a maximum of one out of three nodes crashing. It would not make sense to tolerate any number of faults: if all nodes crash, there is nothing that can be done. 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 space—good luck getting that budget item approved.
Counter-intuitively, in such fault-tolerant systems, it can make sense to *increase* the rate of faults by triggering them deliberately—for example, by randomly killing individual processes without warning. Many critical bugs are actually due to poor error handling [[34](ch02.html#Yuan2014)]; 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. *Chaos engineering* is a discipline that aims to improve confidence in fault-tolerance mechanisms through experiments such as deliberately injecting faults [[35](ch02.html#Rosenthal2020)].
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.
- Approximately 2–5% of magnetic hard drives fail per year [[36](ch02.html#Pinheiro2007), [37](ch02.html#Schroeder2007)]; in a storage cluster with 10,000 disks, we should therefore expect on average one disk failure per day. Recent data suggests that disks are getting more reliable, but failure rates remain significant [[38](ch02.html#Klein2021)].
- Approximately 0.5–1% of solid state drives (SSDs) fail per year [[39](ch02.html#Narayanan2016)]. Small numbers of bit errors are corrected automatically [[40](ch02.html#Alibaba2019_ch2)], but uncorrectable errors occur approximately once per year per drive, even in drives that are fairly new (i.e., that have experienced little wear); this error rate is higher than that of magnetic hard drives [[41](ch02.html#Schroeder2016), [42](ch02.html#Alter2019)].
- Other hardware components such as power supplies, RAID controllers, and memory modules also fail, although less frequently than hard drives [[43](ch02.html#Ford2010), [44](ch02.html#Vishwanath2010)].
- Approximately one in 1,000 machines has a CPU core that occasionally computes the wrong result, likely due to manufacturing defects [[45](ch02.html#Hochschild2021), [46](ch02.html#Dixit2021), [47](ch02.html#Behrens2015)]. In some cases, an erroneous computation leads to a crash, but in other cases it leads to a program simply returning the wrong result.
- Data in RAM can also be corrupted, either due to random events such as cosmic rays, or due to permanent physical defects. Even when memory with error-correcting codes (ECC) is used, more than 1% of machines encounter an uncorrectable error in a given year, which typically leads to a crash of the machine and the affected memory module needing to be replaced [[48](ch02.html#Schroeder2009)]. Moreover, certain pathological memory access patterns can flip bits with high probability [[49](ch02.html#Kim2014)].
- An entire datacenter might become unavailable (for example, due to power outage or network misconfiguration) or even be permanently destroyed (for example by fire or flood). Although such large-scale failures are rare, their impact can be catastrophic if a service cannot tolerate the loss of a datacenter [[50](ch02.html#Cockcroft2019)].
These events are rare enough that you often don’t need to worry about them when working on a small system, as long as you can easily replace hardware that becomes faulty. However, in a large-scale system, hardware faults happen often enough that they become part of the normal system operation.
Our first response to unreliable hardware is usually to add redundancy to the individual hardware components in order to reduce the failure rate of the system. Disks may be set up in a RAID configuration (spreading data across multiple disks in the same machine so that a failed disk does not cause data loss), servers may have dual power supplies and hot-swappable CPUs, and datacenters may have batteries and diesel generators for backup power. Such redundancy can often keep a machine running uninterrupted for years.
Redundancy is most effective when component faults are independent, that is, the occurrence of one fault does not change how likely it is that another fault will occur. However, experience has shown that there are often significant correlations between component failures [[37](ch02.html#Schroeder2007), [51](ch02.html#Han2021), [52](ch02.html#Nightingale2011)]; unavailability of an entire server rack or an entire datacenter still happens more often than we would like.
Hardware redundancy increases the uptime of a single machine; however, as discussed in [“Distributed versus Single-Node Systems”](ch01.html#sec_introduction_distributed), there are advantages to using a distributed system, such as being able to tolerate a complete outage of one datacenter. For this reason, cloud systems tend to focus less on the reliability of individual machines, and instead aim to make services highly available by tolerating faulty nodes at the software level. Cloud providers use *availability zones* to identify which resources are physically co-located; resources in the same place are more likely to fail at the same time than geographically separated resources.
The fault-tolerance techniques we discuss in this book are designed to tolerate the loss of entire machines, racks, or availability zones. They generally work by allowing a machine in one datacenter to take over when a machine in another datacenter fails or becomes unreachable. We will discuss such techniques for fault tolerance in [Link to Come], [Link to Come], and at various other points in this book.
Systems that can tolerate the loss of entire machines 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 multi-node fault-tolerant system can be patched by restarting one node at a time, without affecting the service for users. This is called a *rolling upgrade*, and we will discuss it further in [Link to Come].
Although hardware failures can be weakly correlated, they are still mostly independent: for example, if one disk fails, it’s likely that other disks in the same machine will be fine for another while. On the other hand, software faults are often very highly correlated, because it is common for many nodes to run the same software and thus have the same bugs [[53](ch02.html#Gunawi2014), [54](ch02.html#Kreps2012_ch1)]. Such faults are harder to anticipate, and they tend to cause many more system failures than uncorrelated hardware faults [[43](ch02.html#Ford2010)]. For example:
- A software bug that causes every node to fail at the same time in particular circumstances. For example, on June 30, 2012, a leap second caused many Java applications to hang simultaneously due to a bug in the Linux kernel, bringing down many Internet services [[55](ch02.html#Minar2012_ch1)]. Due to a firmware bug, all SSDs of certain models suddenly fail after precisely 32,768 hours of operation (less than 4 years), rendering the data on them unrecoverable [[56](ch02.html#HPE2019)].
- A runaway process that uses up some shared, limited resource, such as CPU time, memory, disk space, network bandwidth, or threads [[57](ch02.html#Hochstein2020)]. For example, a process that consumes too much memory while processing a large request may be killed by the operating system.
- An interaction between different systems results in emergent behavior that does not occur when each system was tested in isolation [[58](ch02.html#Tang2023)].
- Cascading failures, where a problem in one component causes another component to become overloaded and slow down, which in turn brings down another component [[59](ch02.html#Ulrich2016), [60](ch02.html#Fassbender2022)].
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 environment—and while that assumption is usually true, it eventually stops being true for some reason [[61](ch02.html#Cook2000), [62](ch02.html#Woods2017)].
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; avoiding feedback loops such as retry storms (see [“When an overloaded system won’t recover”](ch02.html#sidebar_metastable)); measuring, monitoring, and analyzing system behavior in production.
Humans design and build software systems, and the operators who keep the systems running are also human. Unlike machines, humans don’t just follow rules; their strength is being creative and adaptive in getting their job done. However, this characteristic also leads to unpredictability, and sometimes mistakes that can lead to failures, despite best intentions. For example, one study of large internet services found that configuration changes by operators were the leading cause of outages, whereas hardware faults (servers or network) played a role in only 10–25% of outages [[63](ch02.html#Oppenheimer2003)].
It is tempting to label such problems as “human error” and to wish that they could be solved by better controlling human behavior through tighter procedures and compliance with rules. However, blaming people for mistakes is counterproductive. What we call “human error” is not really the cause of an incident, but rather a symptom of a problem with the sociotechnical system in which people are trying their best to do their jobs [[64](ch02.html#Dekker2017)].
Various technical measures can help minimize the impact of human mistakes, including thorough testing [[34](ch02.html#Yuan2014)], rollback mechanisms for quickly reverting configuration changes, gradual roll-outs of new code, detailed and clear monitoring, observability tools for diagnosing production issues (see [“Problems with Distributed Systems”](ch01.html#sec_introduction_dist_sys_problems)), and well-designed interfaces that encourage “the right thing” and discourage “the wrong thing”.
However, these things require an investment of time and money, and in the pragmatic reality of everyday business, organizations often prioritize revenue-generating activities over measures that increase their resilience against mistakes. If there is a choice between more features and more testing, many organizations understandably choose features. Given this choice, when a preventable mistake inevitably occurs, it does not make sense to blame the person who made the mistake—the problem is the organization’s priorities.
Increasingly, organizations are adopting a culture of *blameless postmortems*: after an incident, the people involved are encouraged to share full details about what happened, without fear of punishment, since this allows others in the organization to learn how to prevent similar problems in the future [[65](ch02.html#Allspaw2012)]. This process may uncover a need to change business priorities, a need to invest in areas that have been neglected, a need to change the incentives for the people involved, or some other systemic issue that needs to be brought to the management’s attention.
As a general principle, when investigating an incident, you should be suspicious of simplistic answers. “Bob should have been more careful when deploying that change” is not productive, but neither is “We must rewrite the backend in Haskell.” Instead, management should take the opportunity to learn the details of how the sociotechnical system works from the point of view of the people who work with it every day, and take steps to improve it based on this feedback [[64](ch02.html#Dekker2017)].
Reliability is not just for nuclear power stations and air traffic control—more mundane applications are also expected to work reliably. Bugs in business applications cause lost productivity (and legal risks if figures are reported incorrectly), and outages of e-commerce sites can have huge costs in terms of lost revenue and damage to reputation.
In many applications, a temporary outage of a few minutes or even a few hours is tolerable [[66](ch02.html#Sabo2023)], but permanent data loss or corruption would be catastrophic. Consider a parent who stores all their pictures and videos of their children in your photo application [[67](ch02.html#Jurewitz2013)]. How would they feel if that database was suddenly corrupted? Would they know how to restore it from a backup?
As another example of how unreliable software can harm people, consider the Post Office Horizon scandal. Between 1999 and 2019, hundreds of people managing Post Office branches in Britain were convicted of theft or fraud because the accounting software showed a shortfall in their accounts. Eventually it became clear that many of these shortfalls were due to bugs in the software, and many convictions have since been overturned [[68](ch02.html#Siddique2021)]. What led to this, probably the largest miscarriage of justice in British history, is the fact that English law assumes that computers operate correctly (and hence, evidence produced by computers is reliable) unless there is evidence to the contrary [[69](ch02.html#Bohm2022)]. Software engineers may laugh at the idea that software could ever be bug-free, but this is little solace to the people who were wrongfully imprisoned, declared bankrupt, or even committed suicide as a result of a wrongful conviction due to an unreliable computer system.
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 market)—but we should be very conscious of when we are cutting corners and keep in mind the potential consequences.
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. Sometimes, when discussing scalability, people make comments along the lines of, “You’re not Google or Amazon. Stop worrying about scale and just use a relational database.” Whether this maxim applies to you depends on the type of application you are building.
If you are building a new product that currently only has a small number of users, perhaps at a startup, the overriding engineering goal is usually to keep the system as simple and flexible as possible, so that you can easily modify and adapt the features of your product as you learn more about customers’ needs [[70](ch02.html#McKinley2015)]. In such an environment, it is counterproductive to worry about hypothetical scale that might be needed in the future: in the best case, investments in scalability are wasted effort and premature optimization; in the worst case, they lock you into an inflexible design and make it harder to evolve your application.
The reason is that scalability is not a one-dimensional label: it is meaningless to say “X is scalable” or “Y doesn’t scale.” Rather, discussing scalability means considering questions like:
If you succeed in making your application popular, and therefore handling a growing amount of load, you will learn where your performance bottlenecks lie, and therefore you will know along which dimensions you need to scale. At that point it’s time to start worrying about techniques for scalability.
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?). Often this will be a measure of throughput: for example, the number of requests per second to a service, how many gigabytes of new data arrive per day, or the number of shopping cart checkouts per hour. Sometimes you care about the peak of some variable quantity, such as the number of simultaneously online users in [“Case Study: Social Network Home Timelines”](ch02.html#sec_introduction_twitter).
Often there are other statistical characteristics of the load that also affect the access patterns and hence the scalability requirements. For example, you may need to know the ratio of reads to writes in a database, the hit rate on a cache, or the number of data items per user (for example, the number of followers in the social network case study). Perhaps the average case is what matters for you, or perhaps your bottleneck is dominated by a small number of extreme cases. It all depends on the details of your particular application.
- When you increase the load in a certain way and keep the system resources (CPUs, memory, network bandwidth, etc.) unchanged, how is the performance of your system affected?
- When you increase the load in a certain way, how much do you need to increase the resources if you want to keep performance unchanged?
Usually our goal is to keep the performance of the system within the requirements of the SLA (see [“Use of Response Time Metrics”](ch02.html#sec_introduction_slo_sla)) while also minimizing the cost of running the system. The greater the required computing resources, the higher the cost. It might be that some types of hardware are more cost-effective than others, and these factors may change over time as new types of hardware become available.
If you can double the resources in order to handle twice the load, while keeping performance the same, we say that you have *linear scalability*, and this is considered a good thing. Occasionally it is possible to handle twice the load with less than double the resources, due to economies of scale or a better distribution of peak load [[71](ch02.html#Warfield2023), [72](ch02.html#Brooker2023)]. Much more likely is that the cost grows faster than linearly, and there may be many reasons for the inefficiency. For example, if you have a lot of data, then processing a single write request may involve more work than if you have a small amount of data, even if the size of the request is the same.
The simplest way of increasing the hardware resources of a service is to move it to a more powerful machine. Individual CPU cores are no longer getting significantly faster, but you can buy a machine (or rent a cloud instance) with more CPU cores, more RAM, and more disk space. This approach is called *vertical scaling* or *scaling up*.
You can get parallelism on a single machine by using multiple processes or threads. All the threads belonging to the same process can access the same RAM, and hence this approach is also called a *shared-memory architecture*. The problem with a shared-memory approach is that the cost grows faster than linearly: a high-end machine with twice the hardware resources typically costs significantly more than twice as much. And due to bottlenecks, a machine twice the size can often handle less than twice the load.
Another approach is the *shared-disk architecture*, which uses several machines with independent CPUs and RAM, but which stores data on an array of disks that is shared between the machines, which are connected via a fast network: *Network-Attached Storage* (NAS) or *Storage Area Network* (SAN). This architecture has traditionally been used for on-premises data warehousing workloads, but contention and the overhead of locking limit the scalability of the shared-disk approach [[73](ch02.html#Stopford2009)].
By contrast, the *shared-nothing architecture* [[74](ch02.html#Stonebraker1986)] (also called *horizontal scaling* or *scaling out*) has gained a lot of popularity. In this approach, we use a distributed system with multiple nodes, each of which has its own CPUs, RAM, and disks. Any coordination between nodes is done at the software level, via a conventional network.
The advantages of shared-nothing are that it has the potential to scale linearly, it can use whatever hardware offers the best price/performance ratio (especially in the cloud), it can more easily adjust its hardware resources as load increases or decreases, and it can achieve greater fault tolerance by distributing the system across multiple data centers and regions. The downsides are that it requires explicit data partitioning (see [Link to Come]), and it incurs all the complexity of distributed systems ([Link to Come]).
Some cloud-native database systems use separate services for storage and transaction execution (see [“Separation of storage and compute”](ch01.html#sec_introduction_storage_compute)), with multiple compute nodes sharing access to the same storage service. This model has some similarity to a shared-disk architecture, but it avoids the scalability problems of older systems: instead of providing a filesystem (NAS) or block device (SAN) abstraction, the storage service offers a specialized API that is designed for the specific needs of the database [[75](ch02.html#Antonopoulos2019_ch2)].
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*). 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 throughput (100 MB/sec).
Moreover, 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. As the needs of the application are likely to evolve, it is usually not worth planning future scaling needs more than one order of magnitude in advance.
A good general principle for scalability is to break a system down into smaller components that can operate largely independently from each other. This is the underlying principle behind microservices (see [“Microservices and Serverless”](ch01.html#sec_introduction_microservices)), partitioning ([Link to Come]), stream processing ([Link to Come]), and shared-nothing architectures. However, the challenge is in knowing where to draw the line between things that should be together, and things that should be apart. Design guidelines for microservices can be found in other books [[76](ch02.html#Newman2021_ch2)], and we discuss partitioning of shared-nothing systems in [Link to Come].
Another good principle is not to make things more complicated than necessary. If a single-machine database will do the job, it’s probably preferable to a complicated distributed setup. Auto-scaling systems (which automatically add or remove resources in response to demand) are cool, but if your load is fairly predictable, a manually scaled system may have fewer operational surprises (see [Link to Come]). A system with five services is simpler than one with fifty. Good architectures usually involve a pragmatic mixture of approaches.
Software does not wear out or suffer material fatigue, so it does not break in the same ways as mechanical objects do. But the requirements for an application frequently change, the environment that the software runs in changes (such as its dependencies and the underlying platform), and it has bugs that need fixing.
It is widely recognized that the majority of the cost of software is not in its initial development, 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 [[77](ch02.html#Ensmenger2016), [78](ch02.html#Glass2002)].
However, maintenance is also difficult. If a system has been successfully running for a long time, it may well use outdated technologies that not many engineers understand today (such as mainframes and COBOL code); institutional knowledge of how and why a system was designed in a certain way may have been lost as people have left the organization; it might be necessary to fix other people’s mistakes. Moreover, the computer system is often intertwined with the human organization that it supports, which means that maintenance of such *legacy* systems is as much a people problem as a technical one [[79](ch02.html#Bellotti2021)].
Every system we create today will one day become a legacy system if it is valuable enough to survive for a long time. In order to minimize the pain for future generations who need to maintain our software, we should design it with maintenance concerns in mind. Although we cannot always predict which decisions might create maintenance headaches in the future, in this book we will pay attention to several principles that are widely applicable:
We previously discussed the role of operations in [“Operations in the Cloud Era”](ch01.html#sec_introduction_operations), and we saw that human processes are at least as important for reliable operations as software tools. In fact, 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 operations” [[54](ch02.html#Kreps2012_ch1)].
In large-scale systems consisting of many thousands of machines, manual maintenance would be unreasonably expensive, and automation is essential. However, automation can be a two-edged sword: there will always be edge cases (such as rare failure scenarios) that require manual intervention from the operations team. Since the cases that cannot be handled automatically are the most complex issues, greater automation requires a *more* skilled operations team that can resolve those issues [[80](ch02.html#Bainbridge1983)].
Moreover, if an automated system goes wrong, it is often harder to troubleshoot than a system that relies on an operator to perform some actions manually. For that reason, it is not the case that more automation is always better for operability. However, some amount of automation is important, and the sweet spot will depend on the specifics of your particular application and organization.
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 [[81](ch02.html#Hamilton2007)]:
- Allowing monitoring tools to check the system’s key metrics, and supporting observability tools (see [“Problems with Distributed Systems”](ch01.html#sec_introduction_dist_sys_problems)) to give insights into the system’s runtime behavior. A variety of commercial and open source tools can help here [[82](ch02.html#Horovits2021)].
- Avoiding dependency on individual machines (allowing machines to be taken down for maintenance while the system as a whole continues running uninterrupted)
- 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
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 increasing the cost of maintenance. A software project mired in complexity is sometimes described as a *big ball of mud* [[83](ch02.html#Foote1997)].
When complexity makes maintenance hard, budgets and schedules are often overrun. In complex software, there is also a greater risk of introducing bugs when making 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 [[62](ch02.html#Woods2017)]. Conversely, reducing complexity greatly improves the maintainability of software, and thus simplicity should be a key goal for the systems we build.
Simple systems are easier to understand, and therefore we should try to solve a given problem in the simplest way possible. Unfortunately, this is easier said than done. Whether something is simple or not is often a subjective matter of taste, as there is no objective standard of simplicity [[84](ch02.html#Brooker2022)]. For example, one system may hide a complex implementation behind a simple interface, whereas another may have a simple implementation that exposes more internal detail to its users—which one is simpler?
One attempt at reasoning about complexity has been to break it down into two categories, *essential* and *accidental* complexity [[85](ch02.html#Brooks1995)]. The idea is that essential complexity is inherent in the problem domain of the application, while accidental complexity arises only because of limitations of our tooling. Unfortunately, this distinction is also flawed, because boundaries between the essential and the accidental shift as our tooling evolves [[86](ch02.html#Luu2020)].
One of the best tools we have for managing 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 inconsistencies 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.
Abstractions for application code, which aim to reduce its complexity, can be created using methodologies such as *design patterns* [[87](ch02.html#Gamma1994)] and *domain-driven design* (DDD) [[88](ch02.html#Evans2003)]. This book is not about such application-specific abstractions, but rather about general-purpose abstractions on top of which you can build your applications, such as database transactions, indexes, and event logs. If you want to use techniques such as DDD, you can implement them on top of the foundations described in this book.
It’s extremely unlikely that your system’s requirements will remain unchanged forever. They are much more likely to be in constant flux: you learn new facts, previously unanticipated use cases emerge, business priorities change, users request new 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 processes that are helpful when developing software in a frequently changing environment, such as test-driven development (TDD) and refactoring. In this book, we search for ways of increasing agility at the level of a system consisting of several different applications or services with different characteristics.
The ease with which you can modify a data system, and adapt it to changing requirements, is closely linked to its simplicity and its abstractions: simple and easy-to-understand systems are usually easier to modify than complex ones. Since this is such an important idea, we will use a different word to refer to agility on a data system level: *evolvability* [[89](ch02.html#Breivold2008)].
One major factor that makes change difficult in large systems is when some action is irreversible, and therefore that action needs to be taken very carefully [[90](ch02.html#Zaninotto2002)]. For example, say you are migrating from one database to another: if you cannot switch back to the old system in case of problems wth the new one, the stakes are much higher than if you can easily go back. Minimizing irreversibility improves flexibility.
In this chapter we examined several examples of nonfunctional requirements: performance, reliability, scalability, and maintainability. Through these topics we have also encountered principles and terminology that we will need throughout the rest of the book. We started with a case study of how one might implement home timelines in a social network, which illustrated some of the challenges that arise at scale.
We discussed how to measure performance (e.g., using response time percentiles), the load on a system (e.g., using throughput metrics), and how they are used in SLAs. Scalability is a closely related concept: that is, ensuring performance stays the same when the load grows. We saw some general principles for scalability, such as breaking a task down into smaller parts that can operate independently, and we will dive into deep technical detail on scalability techniques in the following chapters.
To achieve reliability, you can use fault tolerance techniques, which allow a system to continue providing its service even if some component (e.g., a disk, a machine, or another service) is faulty. We saw examples of hardware faults that can occur, and distinguished them from software faults, which can be harder to deal with because they are often strongly correlated. Another aspect of achieving reliability is to build resilience against humans making mistakes, and we saw blameless postmortems as a technique for learning from incidents.
Finally, we examined several facets of maintainability, including supporting the work of operations teams, managing complexity, and making it easy to evolve an application’s functionality over time. There are no easy answers on how to achieve these things, but one thing that can help is to build applications using well-understood building blocks that provide useful abstractions. The rest of this book will cover a selection of the most important such building blocks.
[[1](ch02.html#Cvet2016-marker)] Mike Cvet. [How We Learned to Stop Worrying and Love Fan-In at Twitter](https://www.youtube.com/watch?v=WEgCjwyXvwc). At *QCon San Francisco*, December 2016.
[[2](ch02.html#Krikorian2012_ch2-marker)] Raffi Krikorian. [Timelines at Scale](http://www.infoq.com/presentations/Twitter-Timeline-Scalability). At *QCon San Francisco*, November 2012. Archived at [perma.cc/V9G5-KLYK](https://perma.cc/V9G5-KLYK)
[[4](ch02.html#Krikorian2013-marker)] Raffi Krikorian. [New Tweets per second record, and how!](https://blog.twitter.com/engineering/en_us/a/2013/new-tweets-per-second-record-and-how) *blog.twitter.com*, August 2013. Archived at [perma.cc/6JZN-XJYN](https://perma.cc/6JZN-XJYN)
[[5](ch02.html#Axon2010_ch2-marker)] Samuel Axon. [3% of Twitter’s Servers Dedicated to Justin Bieber](http://mashable.com/2010/09/07/justin-bieber-twitter/). *mashable.com*, September 2010. Archived at [perma.cc/F35N-CGVX](https://perma.cc/F35N-CGVX)
[[6](ch02.html#Bronson2021-marker)] Nathan Bronson, Abutalib Aghayev, Aleksey Charapko, and Timothy Zhu. [Metastable Failures in Distributed Systems](https://sigops.org/s/conferences/hotos/2021/papers/hotos21-s11-bronson.pdf). At *Workshop on Hot Topics in Operating Systems* (HotOS), May 2021. [doi:10.1145/3458336.3465286](https://doi.org/10.1145/3458336.3465286)
[[7](ch02.html#Brooker2021-marker)] Marc Brooker. [Metastability and Distributed Systems](https://brooker.co.za/blog/2021/05/24/metastable.html). *brooker.co.za*, May 2021. Archived at [archive.org](https://web.archive.org/web/20230324043015/https://brooker.co.za/blog/2021/05/24/metastable.html)
[[8](ch02.html#Brooker2015-marker)] Marc Brooker. [Exponential Backoff And Jitter](https://aws.amazon.com/blogs/architecture/exponential-backoff-and-jitter/). *aws.amazon.com*, March 2015. Archived at [perma.cc/R6MS-AZKH](https://perma.cc/R6MS-AZKH)
[[9](ch02.html#Brooker2022backoff-marker)] Marc Brooker. [What is Backoff For?](https://brooker.co.za/blog/2022/08/11/backoff.html) *brooker.co.za*, August 2022. Archived at [archive.org](https://web.archive.org/web/20230331022111/https://brooker.co.za/blog/2022/08/11/backoff.html)
[[10](ch02.html#Nygard2018-marker)] Michael T. Nygard. [*Release It!*](https://learning.oreilly.com/library/view/release-it-2nd/9781680504552/), 2nd Edition. Pragmatic Bookshelf, January 2018. ISBN: 9781680502398
[[11](ch02.html#Brooker2022retries-marker)] Marc Brooker. [Fixing retries with token buckets and circuit breakers](https://brooker.co.za/blog/2022/02/28/retries.html). *brooker.co.za*, February 2022. Archived at [archive.org](https://web.archive.org/web/20230325195445/https://brooker.co.za/blog/2022/02/28/retries.html)
[[13](ch02.html#Sackman2016_ch2-marker)] Matthew Sackman. [Pushing Back](https://wellquite.org/posts/lshift/pushing_back/). *wellquite.org*, May 2016. Archived at [perma.cc/3KCZ-RUFY](https://perma.cc/3KCZ-RUFY)
[[14](ch02.html#Kopytkov2018-marker)] Dmitry Kopytkov and Patrick Lee. [Meet Bandaid, the Dropbox service proxy](https://dropbox.tech/infrastructure/meet-bandaid-the-dropbox-service-proxy). *dropbox.tech*, March 2018. Archived at [perma.cc/KUU6-YG4S](https://perma.cc/KUU6-YG4S)
[[15](ch02.html#Gunawi2018-marker)] Haryadi S. Gunawi, Riza O. Suminto, Russell Sears, Casey Golliher, Swaminathan Sundararaman, Xing Lin, Tim Emami, Weiguang Sheng, Nematollah Bidokhti, Caitie McCaffrey, Gary Grider, Parks M. Fields, Kevin Harms, Robert B. Ross, Andree Jacobson, Robert Ricci, Kirk Webb, Peter Alvaro, H. Birali Runesha, Mingzhe Hao, and Huaicheng Li. [Fail-Slow at Scale: Evidence of Hardware Performance Faults in Large Production Systems](https://www.usenix.org/system/files/conference/fast18/fast18-gunawi.pdf). At *16th USENIX Conference on File and Storage Technologies*, February 2018.
[[16](ch02.html#DeCandia2007_ch1-marker)] Giuseppe DeCandia, Deniz Hastorun, Madan Jampani, Gunavardhan Kakulapati, Avinash Lakshman, Alex Pilchin, Swaminathan Sivasubramanian, Peter Vosshall, and Werner Vogels. [Dynamo: Amazon’s Highly Available Key-Value Store](http://www.allthingsdistributed.com/files/amazon-dynamo-sosp2007.pdf). At *21st ACM Symposium on Operating Systems Principles* (SOSP), October 2007. [doi:10.1145/1294261.1294281](https://doi.org/10.1145/1294261.1294281)
[[17](ch02.html#Whitenton2020-marker)] Kathryn Whitenton. [The Need for Speed, 23 Years Later](https://www.nngroup.com/articles/the-need-for-speed/). *nngroup.com*, May 2020. Archived at [perma.cc/C4ER-LZYA](https://perma.cc/C4ER-LZYA)
[[18](ch02.html#Linden2006-marker)] Greg Linden. [Marissa Mayer at Web 2.0](https://glinden.blogspot.com/2006/11/marissa-mayer-at-web-20.html). *glinden.blogspot.com*, November 2005. Archived at [perma.cc/V7EA-3VXB](https://perma.cc/V7EA-3VXB)
[[19](ch02.html#Brutlag2009-marker)] Jake Brutlag. [Speed Matters for Google Web Search](https://services.google.com/fh/files/blogs/google_delayexp.pdf). *services.google.com*, June 2009. Archived at [perma.cc/BK7R-X7M2](https://perma.cc/BK7R-X7M2)
[[20](ch02.html#Schurman2009-marker)] Eric Schurman and Jake Brutlag. [Performance Related Changes and their User Impact](https://www.youtube.com/watch?v=bQSE51-gr2s). Talk at *Velocity 2009*.
[[21](ch02.html#Akamai2017-marker)] Akamai Technologies, Inc. [The State of Online Retail Performance](https://web.archive.org/web/20210729180749/https://www.akamai.com/us/en/multimedia/documents/report/akamai-state-of-online-retail-performance-spring-2017.pdf). *akamai.com*, April 2017. Archived at [perma.cc/UEK2-HYCS](https://perma.cc/UEK2-HYCS)
[[22](ch02.html#Bai2017-marker)] Xiao Bai, Ioannis Arapakis, B. Barla Cambazoglu, and Ana Freire. [Understanding and Leveraging the Impact of Response Latency on User Behaviour in Web Search](https://iarapakis.github.io/papers/TOIS17.pdf). *ACM Transactions on Information Systems*, volume 36, issue 2, article 21, April 2018. [doi:10.1145/3106372](https://doi.org/10.1145/3106372)
[[23](ch02.html#Dean2013-marker)] Jeffrey Dean and Luiz André Barroso. [The Tail at Scale](http://cacm.acm.org/magazines/2013/2/160173-the-tail-at-scale/fulltext). *Communications of the ACM*, volume 56, issue 2, pages 74–80, February 2013. [doi:10.1145/2408776.2408794](https://doi.org/10.1145/2408776.2408794)
[[24](ch02.html#Hidalgo2020-marker)] Alex Hidalgo. [*Implementing Service Level Objectives: A Practical Guide to SLIs, SLOs, and Error Budgets*](https://www.oreilly.com/library/view/implementing-service-level/9781492076803/). O’Reilly Media, September 2020. ISBN: 1492076813
[[25](ch02.html#Mogul2019-marker)] Jeffrey C. Mogul and John Wilkes. [Nines are Not Enough: Meaningful Metrics for Clouds](https://research.google/pubs/pub48033/). At *17th Workshop on Hot Topics in Operating Systems* (HotOS), May 2019. [doi:10.1145/3317550.3321432](https://doi.org/10.1145/3317550.3321432)
[[26](ch02.html#Hauer2020-marker)] Tamás Hauer, Philipp Hoffmann, John Lunney, Dan Ardelean, and Amer Diwan. [Meaningful Availability](https://www.usenix.org/conference/nsdi20/presentation/hauer). At *17th USENIX Symposium on Networked Systems Design and Implementation* (NSDI), February 2020.
[[28](ch02.html#Kohn2021-marker)] David Kohn. [How percentile approximation works (and why it’s more useful than averages)](https://www.timescale.com/blog/how-percentile-approximation-works-and-why-its-more-useful-than-averages/). *timescale.com*, September 2021. Archived at [perma.cc/3PDP-NR8B](https://perma.cc/3PDP-NR8B)
[[29](ch02.html#Hartmann2020-marker)] Heinrich Hartmann and Theo Schlossnagle. [Circllhist — A Log-Linear Histogram Data Structure for IT Infrastructure Monitoring](https://arxiv.org/pdf/2001.06561.pdf). *arxiv.org*, January 2020.
[[30](ch02.html#Masson2019-marker)] Charles Masson, Jee E. Rim, and Homin K. Lee. [DDSketch: A Fast and Fully-Mergeable Quantile Sketch with Relative-Error Guarantees](http://www.vldb.org/pvldb/vol12/p2195-masson.pdf). *Proceedings of the VLDB Endowment*, volume 12, issue 12, pages 2195–2205, August 2019. [doi:10.14778/3352063.3352135](https://doi.org/10.14778/3352063.3352135)
[[31](ch02.html#Schwartz2015-marker)] Baron Schwartz. [Why Percentiles Don’t Work the Way You Think](https://orangematter.solarwinds.com/2016/11/18/why-percentiles-dont-work-the-way-you-think/). *solarwinds.com*, November 2016. Archived at [perma.cc/469T-6UGB](https://perma.cc/469T-6UGB)
[[32](ch02.html#Heimerdinger1992-marker)] Walter L. Heimerdinger and Charles B. Weinstock. [A Conceptual Framework for System Fault Tolerance](https://resources.sei.cmu.edu/asset_files/TechnicalReport/1992_005_001_16112.pdf). Technical Report CMU/SEI-92-TR-033, Software Engineering Institute, Carnegie Mellon University, October 1992. Archived at [perma.cc/GD2V-DMJW](https://perma.cc/GD2V-DMJW)
[[33](ch02.html#Gaertner1999-marker)] Felix C. Gärtner. [Fundamentals of fault-tolerant distributed computing in asynchronous environments](https://dl.acm.org/doi/pdf/10.1145/311531.311532). *ACM Computing Surveys*, volume 31, issue 1, pages 1–26, March 1999. [doi:10.1145/311531.311532](https://doi.org/10.1145/311531.311532)
[[34](ch02.html#Yuan2014-marker)] Ding Yuan, Yu Luo, Xin Zhuang, Guilherme Renna Rodrigues, Xu Zhao, Yongle Zhang, Pranay U. Jain, and Michael Stumm. [Simple Testing Can Prevent Most Critical Failures: An Analysis of Production Failures in Distributed Data-Intensive Systems](https://www.usenix.org/system/files/conference/osdi14/osdi14-paper-yuan.pdf). At *11th USENIX Symposium on Operating Systems Design and Implementation* (OSDI), October 2014.
[[36](ch02.html#Pinheiro2007-marker)] Eduardo Pinheiro, Wolf-Dietrich Weber, and Luiz Andre Barroso. [Failure Trends in a Large Disk Drive Population](https://www.usenix.org/legacy/events/fast07/tech/full_papers/pinheiro/pinheiro_old.pdf). At *5th USENIX Conference on File and Storage Technologies* (FAST), February 2007.
[[37](ch02.html#Schroeder2007-marker)] Bianca Schroeder and Garth A. Gibson. [Disk failures in the real world: What does an MTTF of 1,000,000 hours mean to you?](https://www.usenix.org/legacy/events/fast07/tech/schroeder/schroeder.pdf) At *5th USENIX Conference on File and Storage Technologies* (FAST), February 2007.
[[38](ch02.html#Klein2021-marker)] Andy Klein. [Backblaze Drive Stats for Q2 2021](https://www.backblaze.com/blog/backblaze-drive-stats-for-q2-2021/). *backblaze.com*, August 2021. Archived at [perma.cc/2943-UD5E](https://perma.cc/2943-UD5E)
[[39](ch02.html#Narayanan2016-marker)] Iyswarya Narayanan, Di Wang, Myeongjae Jeon, Bikash Sharma, Laura Caulfield, Anand Sivasubramaniam, Ben Cutler, Jie Liu, Badriddine Khessib, and Kushagra Vaid. [SSD Failures in Datacenters: What? When? and Why?](https://www.microsoft.com/en-us/research/wp-content/uploads/2016/08/a7-narayanan.pdf) At *9th ACM International on Systems and Storage Conference* (SYSTOR), June 2016. [doi:10.1145/2928275.2928278](https://doi.org/10.1145/2928275.2928278)
[[41](ch02.html#Schroeder2016-marker)] Bianca Schroeder, Raghav Lagisetty, and Arif Merchant. [Flash Reliability in Production: The Expected and the Unexpected](https://www.usenix.org/system/files/conference/fast16/fast16-papers-schroeder.pdf). At *14th USENIX Conference on File and Storage Technologies* (FAST), February 2016.
[[42](ch02.html#Alter2019-marker)] Jacob Alter, Ji Xue, Alma Dimnaku, and Evgenia Smirni. [SSD failures in the field: symptoms, causes, and prediction models](https://dl.acm.org/doi/pdf/10.1145/3295500.3356172). At *International Conference for High Performance Computing, Networking, Storage and Analysis* (SC), November 2019. [doi:10.1145/3295500.3356172](https://doi.org/10.1145/3295500.3356172)
[[43](ch02.html#Ford2010-marker)] Daniel Ford, François Labelle, Florentina I. Popovici, Murray Stokely, Van-Anh Truong, Luiz Barroso, Carrie Grimes, and Sean Quinlan. [Availability in Globally Distributed Storage Systems](https://www.usenix.org/legacy/event/osdi10/tech/full_papers/Ford.pdf). At *9th USENIX Symposium on Operating Systems Design and Implementation* (OSDI), October 2010.
[[45](ch02.html#Hochschild2021-marker)] Peter H. Hochschild, Paul Turner, Jeffrey C. Mogul, Rama Govindaraju, Parthasarathy Ranganathan, David E. Culler, and Amin Vahdat. [Cores that don’t count](https://sigops.org/s/conferences/hotos/2021/papers/hotos21-s01-hochschild.pdf). At *Workshop on Hot Topics in Operating Systems* (HotOS), June 2021. [doi:10.1145/3458336.3465297](https://doi.org/10.1145/3458336.3465297)
[[46](ch02.html#Dixit2021-marker)] Harish Dattatraya Dixit, Sneha Pendharkar, Matt Beadon, Chris Mason, Tejasvi Chakravarthy, Bharath Muthiah, and Sriram Sankar. [Silent Data Corruptions at Scale](https://arxiv.org/abs/2102.11245). *arXiv:2102.11245*, February 2021.
[[47](ch02.html#Behrens2015-marker)] Diogo Behrens, Marco Serafini, Sergei Arnautov, Flavio P. Junqueira, and Christof Fetzer. [Scalable Error Isolation for Distributed Systems](https://www.usenix.org/conference/nsdi15/technical-sessions/presentation/behrens). At *12th USENIX Symposium on Networked Systems Design and Implementation* (NSDI), May 2015.
[[48](ch02.html#Schroeder2009-marker)] Bianca Schroeder, Eduardo Pinheiro, and Wolf-Dietrich Weber. [DRAM Errors in the Wild: A Large-Scale Field Study](https://static.googleusercontent.com/media/research.google.com/en//pubs/archive/35162.pdf). At *11th International Joint Conference on Measurement and Modeling of Computer Systems* (SIGMETRICS), June 2009. [doi:10.1145/1555349.1555372](https://doi.org/10.1145/1555349.1555372)
[[49](ch02.html#Kim2014-marker)] Yoongu Kim, Ross Daly, Jeremie Kim, Chris Fallin, Ji Hye Lee, Donghyuk Lee, Chris Wilkerson, Konrad Lai, and Onur Mutlu. [Flipping Bits in Memory Without Accessing Them: An Experimental Study of DRAM Disturbance Errors](https://users.ece.cmu.edu/~yoonguk/papers/kim-isca14.pdf). At *41st Annual International Symposium on Computer Architecture* (ISCA), June 2014. [doi:10.5555/2665671.2665726](https://doi.org/10.5555/2665671.2665726)
[[50](ch02.html#Cockcroft2019-marker)] Adrian Cockcroft. [Failure Modes and Continuous Resilience](https://adrianco.medium.com/failure-modes-and-continuous-resilience-6553078caad5). *adrianco.medium.com*, November 2019. Archived at [perma.cc/7SYS-BVJP](https://perma.cc/7SYS-BVJP)
[[51](ch02.html#Han2021-marker)] Shujie Han, Patrick P. C. Lee, Fan Xu, Yi Liu, Cheng He, and Jiongzhou Liu. [An In-Depth Study of Correlated Failures in Production SSD-Based Data Centers](https://www.usenix.org/conference/fast21/presentation/han). At *19th USENIX Conference on File and Storage Technologies* (FAST), February 2021.
[[52](ch02.html#Nightingale2011-marker)] Edmund B. Nightingale, John R. Douceur, and Vince Orgovan. [Cycles, Cells and Platters: An Empirical Analysis of Hardware Failures on a Million Consumer PCs](https://eurosys2011.cs.uni-salzburg.at/pdf/eurosys2011-nightingale.pdf). At *6th European Conference on Computer Systems* (EuroSys), April 2011. [doi:10.1145/1966445.1966477](https://doi.org/10.1145/1966445.1966477)
[[53](ch02.html#Gunawi2014-marker)] Haryadi S. Gunawi, Mingzhe Hao, Tanakorn Leesatapornwongsa, Tiratat Patana-anake, Thanh Do, Jeffry Adityatama, Kurnia J. Eliazar, Agung Laksono, Jeffrey F. Lukman, Vincentius Martin, and Anang D. Satria. [What Bugs Live in the Cloud?](http://ucare.cs.uchicago.edu/pdf/socc14-cbs.pdf) At *5th ACM Symposium on Cloud Computing* (SoCC), November 2014. [doi:10.1145/2670979.2670986](https://doi.org/10.1145/2670979.2670986)
[[54](ch02.html#Kreps2012_ch1-marker)] Jay Kreps. [Getting Real About Distributed System Reliability](http://blog.empathybox.com/post/19574936361/getting-real-about-distributed-system-reliability). *blog.empathybox.com*, March 2012. Archived at [perma.cc/9B5Q-AEBW](https://perma.cc/9B5Q-AEBW)
[[55](ch02.html#Minar2012_ch1-marker)] Nelson Minar. [Leap Second Crashes Half the Internet](http://www.somebits.com/weblog/tech/bad/leap-second-2012.html). *somebits.com*, July 2012. Archived at [perma.cc/2WB8-D6EU](https://perma.cc/2WB8-D6EU)
[[58](ch02.html#Tang2023-marker)] Lilia Tang, Chaitanya Bhandari, Yongle Zhang, Anna Karanika, Shuyang Ji, Indranil Gupta, and Tianyin Xu. [Fail through the Cracks: Cross-System Interaction Failures in Modern Cloud Systems](https://tianyin.github.io/pub/csi-failures.pdf). At *18th European Conference on Computer Systems* (EuroSys), May 2023. [doi:10.1145/3552326.3587448](https://doi.org/10.1145/3552326.3587448)
[[59](ch02.html#Ulrich2016-marker)] Mike Ulrich. [Addressing Cascading Failures](https://sre.google/sre-book/addressing-cascading-failures/). In Betsy Beyer, Jennifer Petoff, Chris Jones, and Niall Richard Murphy (ed). [*Site Reliability Engineering: How Google Runs Production Systems*](https://www.oreilly.com/library/view/site-reliability-engineering/9781491929117/). O’Reilly Media, 2016. ISBN: 9781491929124
[[60](ch02.html#Fassbender2022-marker)] Harri Faßbender. [Cascading failures in large-scale distributed systems](https://blog.mi.hdm-stuttgart.de/index.php/2022/03/03/cascading-failures-in-large-scale-distributed-systems/). *blog.mi.hdm-stuttgart.de*, March 2022. Archived at [perma.cc/K7VY-YJRX](https://perma.cc/K7VY-YJRX)
[[61](ch02.html#Cook2000-marker)] Richard I. Cook. [How Complex Systems Fail](https://www.adaptivecapacitylabs.com/HowComplexSystemsFail.pdf). Cognitive Technologies Laboratory, April 2000. Archived at [perma.cc/RDS6-2YVA](https://perma.cc/RDS6-2YVA)
[[62](ch02.html#Woods2017-marker)] David D Woods. [STELLA: Report from the SNAFUcatchers Workshop on Coping With Complexity](https://snafucatchers.github.io/). *snafucatchers.github.io*, March 2017. Archived at [archive.org](https://web.archive.org/web/20230306130131/https://snafucatchers.github.io/)
[[63](ch02.html#Oppenheimer2003-marker)] David Oppenheimer, Archana Ganapathi, and David A. Patterson. [Why Do Internet Services Fail, and What Can Be Done About It?](http://static.usenix.org/legacy/events/usits03/tech/full_papers/oppenheimer/oppenheimer.pdf) At *4th USENIX Symposium on Internet Technologies and Systems* (USITS), March 2003.
[[65](ch02.html#Allspaw2012-marker)] John Allspaw. [Blameless PostMortems and a Just Culture](https://www.etsy.com/codeascraft/blameless-postmortems/). *etsy.com*, May 2012. Archived at [perma.cc/YMJ7-NTAP](https://perma.cc/YMJ7-NTAP)
[[66](ch02.html#Sabo2023-marker)] Itzy Sabo. [Uptime Guarantees — A Pragmatic Perspective](https://world.hey.com/itzy/uptime-guarantees-a-pragmatic-perspective-736d7ea4). *world.hey.com*, March 2023. Archived at [perma.cc/F7TU-78JB](https://perma.cc/F7TU-78JB)
[[67](ch02.html#Jurewitz2013-marker)] Michael Jurewitz. [The Human Impact of Bugs](http://jury.me/blog/2013/3/14/the-human-impact-of-bugs). *jury.me*, March 2013. Archived at [perma.cc/5KQ4-VDYL](https://perma.cc/5KQ4-VDYL)
[[68](ch02.html#Siddique2021-marker)] Haroon Siddique and Ben Quinn. [Court clears 39 post office operators convicted due to ‘corrupt data’](https://www.theguardian.com/uk-news/2021/apr/23/court-clears-39-post-office-staff-convicted-due-to-corrupt-data). *theguardian.com*, April 2021. Archived at [archive.org](https://web.archive.org/web/20220630124107/https://www.theguardian.com/uk-news/2021/apr/23/court-clears-39-post-office-staff-convicted-due-to-corrupt-data)
[[69](ch02.html#Bohm2022-marker)] Nicholas Bohm, James Christie, Peter Bernard Ladkin, Bev Littlewood, Paul Marshall, Stephen Mason, Martin Newby, Steven J. Murdoch, Harold Thimbleby, and Martyn Thomas. [The legal rule that computers are presumed to be operating correctly – unforeseen and unjust consequences](https://www.benthamsgaze.org/wp-content/uploads/2022/06/briefing-presumption-that-computers-are-reliable.pdf). Briefing note, *benthamsgaze.org*, June 2022. Archived at [perma.cc/WQ6X-TMW4](https://perma.cc/WQ6X-TMW4)
[[70](ch02.html#McKinley2015-marker)] Dan McKinley. [Choose Boring Technology](https://mcfunley.com/choose-boring-technology). *mcfunley.com*, March 2015. Archived at [perma.cc/7QW7-J4YP](https://perma.cc/7QW7-J4YP)
[[71](ch02.html#Warfield2023-marker)] Andy Warfield. [Building and operating a pretty big storage system called S3](https://www.allthingsdistributed.com/2023/07/building-and-operating-a-pretty-big-storage-system.html). *allthingsdistributed.com*, July 2023. Archived at [perma.cc/7LPK-TP7V](https://perma.cc/7LPK-TP7V)
[[72](ch02.html#Brooker2023-marker)] Marc Brooker. [Surprising Scalability of Multitenancy](https://brooker.co.za/blog/2023/03/23/economics.html). *brooker.co.za*, March 2023. Archived at [archive.org](https://web.archive.org/web/20230404065818/https://brooker.co.za/blog/2023/03/23/economics.html)
[[73](ch02.html#Stopford2009-marker)] Ben Stopford. [Shared Nothing vs. Shared Disk Architectures: An Independent View](http://www.benstopford.com/2009/11/24/understanding-the-shared-nothing-architecture/). *benstopford.com*, November 2009. Archived at [perma.cc/7BXH-EDUR](https://perma.cc/7BXH-EDUR)
[[74](ch02.html#Stonebraker1986-marker)] Michael Stonebraker. [The Case for Shared Nothing](http://db.cs.berkeley.edu/papers/hpts85-nothing.pdf). *IEEE Database Engineering Bulletin*, volume 9, issue 1, pages 4–9, March 1986.
[[75](ch02.html#Antonopoulos2019_ch2-marker)] Panagiotis Antonopoulos, Alex Budovski, Cristian Diaconu, Alejandro Hernandez Saenz, Jack Hu, Hanuma Kodavalla, Donald Kossmann, Sandeep Lingam, Umar Farooq Minhas, Naveen Prakash, Vijendra Purohit, Hugh Qu, Chaitanya Sreenivas Ravella, Krystyna Reisteter, Sheetal Shrotri, Dixin Tang, and Vikram Wakade. [Socrates: The New SQL Server in the Cloud](https://www.microsoft.com/en-us/research/uploads/prod/2019/05/socrates.pdf). At *ACM International Conference on Management of Data* (SIGMOD), pages 1743–1756, June 2019. [doi:10.1145/3299869.3314047](https://doi.org/10.1145/3299869.3314047)
[[77](ch02.html#Ensmenger2016-marker)] Nathan Ensmenger. [When Good Software Goes Bad: The Surprising Durability of an Ephemeral Technology](https://themaintainers.wpengine.com/wp-content/uploads/2021/04/ensmenger-maintainers-v2.pdf). At *The Maintainers Conference*, April 2016. Archived at [perma.cc/ZXT4-HGZB](https://perma.cc/ZXT4-HGZB)
[[78](ch02.html#Glass2002-marker)] Robert L. Glass. [*Facts and Fallacies of Software Engineering*](https://learning.oreilly.com/library/view/facts-and-fallacies/0321117425/). Addison-Wesley Professional, October 2002. ISBN: 9780321117427
[[79](ch02.html#Bellotti2021-marker)] Marianne Bellotti. [*Kill It with Fire*](https://learning.oreilly.com/library/view/kill-it-with/9781098128883/). No Starch Press, April 2021. ISBN: 9781718501188
[[81](ch02.html#Hamilton2007-marker)] James Hamilton. [On Designing and Deploying Internet-Scale Services](https://www.usenix.org/legacy/events/lisa07/tech/full_papers/hamilton/hamilton.pdf). At *21st Large Installation System Administration Conference* (LISA), November 2007.
[[82](ch02.html#Horovits2021-marker)] Dotan Horovits. [Open Source for Better Observability](https://horovits.medium.com/open-source-for-better-observability-8c65b5630561). *horovits.medium.com*, October 2021. Archived at [perma.cc/R2HD-U2ZT](https://perma.cc/R2HD-U2ZT)
[[83](ch02.html#Foote1997-marker)] Brian Foote and Joseph Yoder. [Big Ball of Mud](http://www.laputan.org/pub/foote/mud.pdf). At *4th Conference on Pattern Languages of Programs* (PLoP), September 1997. Archived at [perma.cc/4GUP-2PBV](https://perma.cc/4GUP-2PBV)
[[84](ch02.html#Brooker2022-marker)] Marc Brooker. [What is a simple system?](https://brooker.co.za/blog/2022/05/03/simplicity.html) *brooker.co.za*, May 2022. Archived at [archive.org](https://web.archive.org/web/20220602141902/https://brooker.co.za/blog/2022/05/03/simplicity.html)
[[85](ch02.html#Brooks1995-marker)] Frederick P Brooks. [No Silver Bullet – Essence and Accident in Software Engineering](http://worrydream.com/refs/Brooks-NoSilverBullet.pdf). In [*The Mythical Man-Month*](https://www.oreilly.com/library/view/mythical-man-month-the/0201835959/), Anniversary edition, Addison-Wesley, 1995. ISBN: 9780201835953
[[86](ch02.html#Luu2020-marker)] Dan Luu. [Against essential and accidental complexity](https://danluu.com/essential-complexity/). *danluu.com*, December 2020. Archived at [perma.cc/H5ES-69KC](https://perma.cc/H5ES-69KC)
[[87](ch02.html#Gamma1994-marker)] Erich Gamma, Richard Helm, Ralph Johnson, and John Vlissides. [*Design Patterns: Elements of Reusable Object-Oriented Software*](https://learning.oreilly.com/library/view/design-patterns-elements/0201633612/). Addison-Wesley Professional, October 1994. ISBN: 9780201633610
[[88](ch02.html#Evans2003-marker)] Eric Evans. [*Domain-Driven Design: Tackling Complexity in the Heart of Software*](https://learning.oreilly.com/library/view/domain-driven-design-tackling/0321125215/). Addison-Wesley Professional, August 2003. ISBN: 9780321125217
[[89](ch02.html#Breivold2008-marker)] Hongyu Pei Breivold, Ivica Crnkovic, and Peter J. Eriksson. [Analyzing Software Evolvability](http://www.es.mdh.se/pdf_publications/1251.pdf). at *32nd Annual IEEE International Computer Software and Applications Conference* (COMPSAC), July 2008. [doi:10.1109/COMPSAC.2008.50](https://doi.org/10.1109/COMPSAC.2008.50)
[[90](ch02.html#Zaninotto2002-marker)] Enrico Zaninotto. [From X programming to the X organisation](https://martinfowler.com/articles/zaninotto.pdf). At *XP Conference*, May 2002. Archived at [perma.cc/R9AR-QCKZ](https://perma.cc/R9AR-QCKZ)
