- About This Book
- Trying to Define the Cloud
- Moving to AWS
- Essential Characteristics of AWS Cloud Computing
- Operational Benefits of AWS
- Cloud Provider Limitations
- Data Security at AWS
- Network Security at AWS
- Application Security at AWS
- Compliance in the AWS Cloud
- Migrating Applications
- The Well-Architected Framework
- The Well-Architected Tool
- In Conclusion
Essential Characteristics of AWS Cloud Computing
If you haven’t heard of National Institute of Standards and Technology (NIST), a branch of the U.S. government, you’re not alone. Around 2010, NIST began documenting the public cloud. After talking to all the major vendors, it released an initial report in June 2011 defining many cloud components that were common across all the public cloud vendors. The report’s genius was in defining what the emerging public cloud actually was (the command components). Over the years, NIST’s cloud definitions have moved from definitions to becoming standards for how many companies view working in the public cloud. According to NIST, five key definitions of the public cloud have really morphed into a definitive standard methodology of operating in the public cloud:
On-demand self-service—We not only expect cloud service to be delivered quickly; we demand it. All cloud providers offer a self-serve portal as AWS does, as shown in Figure 1-6. You request a cloud service, and in seconds it’s available in your AWS account ready to configure. Gone are the days of requesting a virtual server via email and waiting several days until it’s built. At AWS, a virtual server can be started and operational in seconds. Procuring a software-defined network at AWS (called a virtual private cloud) is available and operational in seconds. AWS has an expansive self-serve management console that allows you to order and configure many cloud-hosted services in seconds in any AWS region. Any cloud service that you order from AWS is automatically delivered to you through heavily automated procedures. There are no public cloud providers that survive without a self-service portal driven by heavy-duty automation in the background. This NIST definition is now a standard.
Figure 1-6 The AWS management portal
Broad network access—Cloud services can be accessed from almost anywhere across the globe using the Internet. If you host applications at AWS, perhaps they are public-facing SaaS apps. AWS also provides HTTPS endpoints to access every cloud service hosted at AWS. However, you may not want broad network access, which is defined as public network access to your cloud services. In fact, many companies that are moving to the AWS cloud have no interest in a publicly accessible software solution. They want their hosted cloud services to remain private, accessible only by their employees using private connections. Each cloud customer ultimately defines the real meaning of broad network access. At AWS, applications can be publicly available, or, you can stay completely private. VPN connections from your place of work to AWS are commonplace; in fact, you can order Direct Connect and establish a private fiber connection to AWS running at speeds up to 10 Gbps. Depending on the type of applications you’re using in the cloud, high-speed network access is essential. We can even use, access, and administer AWS service from our phone using AWS apps. Certainly, accessing AWS from any device is possible. For more details on networking, check out Chapter 3, “AWS Networking Services.”
Resource Pooling—Infrastructure resources for public cloud providers are pooled together in many data centers across the different regions of the world and are dynamically assigned on demand. A company running an on-premise private cloud would pool its virtual machines, memory, processing, and networking capabilities into one or two data centers, and from its own pool offer limited compute resources. All public cloud providers have a massive pool of resources to serve our various needs. AWS has clusters of data centers (known as AZs or availability zones), and each AZ could have over 80,000 bare-metal servers available and online allowing customers to host their application services with a high level of resiliency and failover. Having many available online resources also enables AWS to keep the price down. Without a massive pool of resources, AWS would not be able to offer its cloud services on demand that are able to scale up and down based on customer demand. Having a massive resource pool is a necessary standard for all public cloud providers; customers do not expect to run out of resources. Take, for example, AWS S3 storage, which is unlimited with no defined maximum limit. For more details on regions and AZs, check out Chapter 2.
Rapid Elasticity—Elasticity in the public cloud, or scaling, is the key feature required by all hosted cloud applications. Elasticity at AWS is utilized for both compute and storage. Because most services and applications are built on compute and storage, applications in the AWS cloud have the capability to automatically scale, as shown in Figure 1-7. And elasticity, or scaling, is only useful if it’s automated based on demand. Turning off a virtual server, adding RAM, and turning it back on is not the elasticity that we are interested in; we want horizontal scale—that is, more application servers—not just a bigger server. Real-time monitoring of a hosted cloud application at AWS allows us to react almost instantaneously before the application’s performance is close to degrading. With EC2 Auto Scaling in the background, additional computer resources are automatically ordered and delivered to the application server’s cluster, maintaining the application’s performance. Rapid elasticity based on demand is only possible with real-time monitoring driving automated scale. This is why the public cloud is so popular; with a massive pool of available cloud resources and the ability to automatically scale applications out and in based on demand, at AWS anybody can easily scale application stacks up and down. For more details on deploying scale and elasticity with EC2 Auto Scale, check out Chapter 5, “Planning for Scale and Resiliency.”
Figure 1-7 Applications can scale based on demand in the public cloud
Measured Service—In the cloud, you are only billed for what you use; that’s defined as a measured service. Cloud providers make their money by charging for everything that you use in their data centers, including data transfer costs. Packet flow inbound to the public cloud is usually free; outbound packet flow, or traffic between subnets hosted in different data centers, is usually charged an outbound data transfer fee. Charges are per second, or per minute in the case of computer services like AWS EC2 compute instances, or they are per gigabyte per month in the case of storage services like S3 or virtual hard drives, which at AWS are called elastic block storage (EBS). AWS charges can be broken down into compute, storage, and data transfer charges. If an AWS service is on, the meter is running. Cost management is one of your most important jobs when operating in the cloud. AWS has many useful tools to help you control your costs, including the AWS Simple Pricing Calculator, AWS Budgets, and the Cost Explorer, as shown in Figure 1-8. You can find details on these features in Chapter 2. Being billed for consuming cloud services is a reality that we are all used to. What you also may have to get used to is exactly how you are being billed. Again, you must understand and carefully monitor compute, storage, and data transfer costs. For example, you can order a load balancer at AWS for $30 per month. However, there is an additional charge to be aware of: all the data packets transferred through the load balancer are charged, and that by itself can be a hefty price.
Figure 1-8 AWS Budgets and Cost Explorer track and alert when costs are over budget