7th. Performance and Reliability models

Performance, scalability, and reliability are important features of all enterprise applications. Although there are several ways to improve performance and reliability, this pattern cluster emphasizes how to combine systems that provide services to any number of applications or users for better scalability and availability. The patterns in this chapter provide the basis for effectively adapting to changes in load and peak traffic and improving availability.

Meet Operational requirements

Today's enterprise applications must meet increasing operational requirements, including increased availability, improved performance, and the ability to meet these requirements as the application load increases. This requires the design of the application and supporting infrastructure to achieve maximum scalability and availability.

Scalability

Scalability refers to the ability of one or more systems to handle increasing demands while maintaining an acceptable level of performance. To effectively scale a system, you must determine the nature of the increasing demand and understand its impact on individual system components. After you determine the restrictive components, you can scale up or scale out.

Scale up

Scaling up policies are the ability to increase the processing load of resources (such as processors, memory, or storage) in a single server. For example, you can scale up a database cluster built on a multiprocessor operating system by increasing the number of processors or memory.

Scale out

Scaling out policies increases the capacity of the infrastructure-level processing load by increasing the number of servers (thereby increasing the overall capacity of these servers).

Availability of

Availability refers to the measurable uptime of servers, software or applications running on the server, and relies on careful design and operational principles, including change control, rigorous testing, and rapid upgrades and fallback mechanisms. In order to achieve the highest level of availability, it is important to find all the fault points in the design. A highly available design process should analyze each component in the architecture to verify that the overall performance of the system does not depend on any single piece of hardware or software that performs a specific function or provides access to a specific block of information.

Mode overview

This pattern cluster begins with the server clustering (cluster of servers), which emphasizes the use of server clusters to design infrastructure levels to meet specific availability and scalability requirements. A server cluster is two or more servers that are interconnected to form a unified virtual computing resource.

Server clusters increase the availability of the system by ensuring that another server in the cluster can afford the workload when one server becomes unavailable due to a failure or scheduled shutdown, ensuring that the application remains available to the user. The cluster also enhances scalability by supporting more users while maintaining the current performance level, or by improving the application performance of the current user. Increasing scalability through server clusters can also increase server redundancy, which helps improve system availability, as described above.

The Server clustering mode emphasizes that clustering is a universal design technique that will be applied to two other design modes: load-balanced Cluster (Load Balancing cluster) and failover Cluster (failover clustering). Figure 1 shows the performance and reliability mode cluster.

Figure 1 Performance and Reliability mode cluster

load-balanced Cluster

The load-balanced Cluster mode solves how to maintain acceptable performance by designing and implementing a scalable infrastructure level. This pattern describes a common method of implementing the following features: Balancing incoming Internet Protocol (IP) traffic through a set of read-only or application servers or these farms.

The load Balancing feature distributes requests across all working servers in the farm (see server farm), thereby enhancing the infrastructure-level performance, availability, and scalability of servers including WEB servers, streaming media servers, and virtual private network (VPN) servers. When a server in a load-balanced cluster fails, the load-balancing server also provides failover functionality by re-distributing the load to the remaining servers.

Failover Cluster

The Failover Cluster mode helps design a highly available application infrastructure level to prevent loss of service in the event of a single server failure or software hosting it as a failure. This pattern describes failover clusters and how they provide high availability for read/write storage such as databases, messaging systems, and file and print services.

In a failover cluster, if one server becomes unavailable, another server takes over its tasks and continues to provide services to the end user, a process known as failover. When failover occurs, the user can continue to use the application and will not be aware that another server is serving.

Performance and Reliability modes

Table 1 lists the modes in the performance and Reliability mode cluster. These patterns are arranged in a later pattern that is built on the earlier schema. This means that the direction of development is to evolve from more general patterns (such as Server clustering) to more specific patterns (such as Failover Cluster).

7th. Performance and Reliability models