Ralph Retter’s research while affiliated with T-Systems International GmbH and other places

IT applications and physical businesses often face similar challenges. Customers have to be served quickly; throughput and availability should increase. Concepts such as redundancy and parallelism are inherent in the architectural design of both worlds. However, the complexity of IT systems can hinder them from following architectural principles and design rules to obtain highly scalable and fault-resilient applications. The authors cover the architectural design phases of a cloud application and describe common best practices relevant in each phase. They use a coffee shop as a real-world analogy to avoid IT complexity.

In this chapter, we introduce the fundamentals required for the understanding of the following chapters. As stated in the introduction, the cloud computing properties – access via network, on-demand self-service, measured service (pay-per-use), resource pooling and rapid elasticity – fundamentally change how IT resources are provided and used. It is important to understand why cloud offerings have these properties, how these properties are delivered on different levels of a typical application stack and under which conditions an application benefits from them. We begin by examining application workloads (Sect. 2.2) and show how they influence the decision for the adoption of cloud offerings. Especially, we discuss how applications experiencing different types of workloads can benefit from the cloud computing properties covered in Chap. 1. As in that previous chapter, we use the NIST cloud definition [3] and emphasis on those aspects that are important to understand the following chapters.

In this chapter, the different cloud offerings found in clouds are covered regarding the functionality they provide to customers and the behavior they display. After the overview and general discussion of the impact of cloud computing properties (see Sect. 1. 1 Page 3 in Chap. 1) on offering behavior, we describe different cloud environments (Sect. 3.3) as patterns. These patterns characterize the environments created in different cloud deployment models (see Sect. 2. 4 on Page 54 in Chap. 2) in more detail. Especially, they give an overview of common combinations of the other cloud offering patterns to form an IaaS (41) or PaaS (44) cloud. In the remaining sections of this chapter, we cover cloud offerings combined to provide IaaS or PaaS individually and differentiate between three general functionality-related offering types: processing offerings, storage offerings, and communication offerings.

This chapter covers architectural patterns that describe how applications have to be designed to benefit from a cloud environment. Additionally, it is described how applications themselves can be offered as configurable cloud services. Having introduced cloud service models (see Sect. 2. 3 on Page 39 in Chap. 2) and cloud deployment types (see Sect. 2. 4 on Page 54 in Chap. 2), this chapter describes patterns that architects and developers can use to build cloud-native applications, i.e., applications that display the cloud application properties introduced in Sect. 1. 2 on Page 5 in Chap. 1. Following the overview, fundamental application architectural patterns cover the architectural principles found in most cloud-native applications to enable the cloud application properties. Application component patterns then specify patterns on how to design and build individual components of a cloud-native application, so that the overall application can be built on top of an elastic infrastructure (78) or elastic platform (81). Multi-tenancy patterns describe how cloud applications and individual components can be shared by multiple customers, so called tenants, on different levels of the application stack. Cloud integration patterns finally describe mechanisms on how to integrate multiple cloud environments or cloud environments and on-premise datacenters as well as applications both in and outside the cloud.

This chapter covers architectural patterns that describe how cloud applications as described in Chap. 4, can be managed automatically by separate components (Fig. 5.1). These management components (Sect. 5.2) handle the automated management of cloud-native applications regarding dynamic elasticity, resiliency, updates etc. Due to the pay-per-use property of cloud applications covered in Sect. 1. 1, scaling tasks should be automated, because the number of provisioned IT resources, i.e., the number of provisioned virtual servers, the size of booked storage or the number of application component instances directly affects the runtime costs of an application. Furthermore, environment-based availability (88) assurances, where individual cloud resources can fail at any time, or a node-based availability (85) that does not meet requirements of an application, as well as network partitions, may create the need to monitor applications and automatically react to resource failures.

In Chap. 1, we introduced the basic principles of cloud computing, on-demand self-service, broad network access, pay-per-use, resource pooling and rapid elasticity. In Chaps. 2 and 3, we used a pattern format to describe workloads experienced by cloud applications, the hosting environments they use, and the cloud-specific properties of different cloud offerings in an abstract, vendor-neutral view. Chapter 4 covered patterns on how to deal with these properties in application architectures followed by best practices for managing cloud applications in Chap. 5. Chapter 6 covered compositions of the patterns described in previous chapters to create cloud applications.

This chapter contains patterns that compose patterns covered in previous sections to describe cloud applications. Furthermore, it describes possible distribution scenarios of the composed patterns among different cloud environments forming a hybrid cloud (67) (Fig. 6.1).

The properties of clouds -- elasticity, pay-per-use, and standardization of the runtime infrastructure -- enable cloud providers and users alike to benefit from economies of scale, faster provisioning times, and reduced runtime costs. However, to achieve these benefits, application architects and developers have to respect the characteristics of the cloud environment. To reduce the complexity of cloud application architectures, we propose a pattern-based approach for cloud application design and development. We defined a pattern format to describe the principles of cloud computing, available cloud offerings, and cloud application architectures. Based on this format we developed an architectural pattern language of cloud-based applications: through interrelation of patterns for cloud offering descriptions and cloud application architectures, developers are guided during the identification of cloud environments and architecture patterns applicable to their problems. We cover the proceeding how we identified patterns in various information sources and existing productively used applications, give an overview of previously discovered patterns, and introduce one new pattern. Further, we propose a framework for the organizations of patterns and the guidance of developers during pattern instantiation.