The concept of a Service-Oriented Architecture (SOA) is based on applying the appropriate services to applications, processors, and storage to maximize efficiency and flexibility for the IT environment. Because the term “service-oriented” has been in existence for some time, it has been used in different contexts and for different purposes. One constant throughout its existence has been that it represents a distinct approach for separating concerns. This means that the logic required to solve a large problem can be better constructed, managed, and carried out if broken down into a collection of smaller, related pieces. Each of these pieces addresses a concern or specific part of a problem.
SOA encourages individual units of logic to autonomously exist, yet not be isolated from each other. These units of logic are still required to conform to a set of principles that let them independently evolve, while still maintaining sufficient amounts of commonality and standardization.
In SOA, these units of logic are known as services, which are essentially just repeatable business-oriented tasks. The strengths of the System z platform, combined with SOA, represent a powerful solution for any business that desires the ability to move within its marketplace while retaining all the valuable attributes that define the System z platform.
Fulfilling requirements of the SOA concept requires intelligent processing, operating systems, and storage and an infrastructure that supports its objectives. This “data center fabric” should support and provide the services to both the host and storage that allow data flows to be managed as requirements change. A data center fabric that provides different and optional services is termed a “Service-Oriented Infrastructure (SOI)” and provides administrators the flexibility to use adaptive networking that includes Quality of Service (QoS), data mobility, encryption, and virtual connectivity for fast and reliable deployments, independent of the protocol. This article examines how users should plan for today’s infrastructure needs while building an SOI for the future.
The System z and SOA/SOI
System z data centers have been evolving for decades and have re-established themselves as core processing facilities that contain much more than just legacy business applications. For many years, the mainframe has epitomized reliability, availability, security, and scalability. These strengths are the reason why so many enterprises made the mainframe their deployment platform of choice for their mission-critical applications. The banking industry is a great example of an industry that has a strong affinity for the mainframe. However, mainframe capabilities are becoming increasingly important for all industries, resulting in the mainframe becoming increasingly prevalent in data centers for all industries.
Applications, not the network, should control access to and management of their data. Unfortunately, it becomes more difficult for applications to access their data as new protocols and virtualization techniques have been added in the data center. Enterprises actually deploy and manage many physical networks: separate server-to-storage (SANs), server-to-server (clusters), and storage-to-storage replication networks are common within and between data centers.
Although many organizations have continued to isolate different operating systems and technologies, mainframe data center fabrics have steadily moved from a single switching function to a fully integrated infrastructure that provides a much broader solution for storage interconnectivity. With the emergence of Linux on System z and Node Port ID Virtualization (NPIV), the System z9 and System z10 processors have emerged as multi-protocol devices that support mixed I/O and traffic types, and today can support both FICON and Fibre Channel protocols for simultaneous storage connectivity. This functionality is known as Protocol Intermix Mode (PIM).
PIM, a feature supported by IBM on System z processors, lets FICON and open systems Fibre Channel Protocol (FCP) traffic coexist on the same physical storage network. The enhancements made to the IBM System z family have made PIM a more attractive, viable, and realistic option for connectivity. NPIV technology finally makes it realistic to run open systems and z/OS on a mainframe connecting everything via FICON Express channel cards onto a common storage network. Virtual fabric technologies enable the user to bring UNIX and Windows servers into this storage network.
Much work and vendor cooperation have fueled the development of standards to support these storage networking virtualization techniques, standardized in T11 to the System z9 (and now the System z10) to replace multiple open systems servers with Linux on System z. NPIV lets these servers use open systems storage for a low-cost solution with enhanced technologies. The Fibre Channel virtual switches that create virtual fabrics are another capability to increase manageability and let physical switches be more adaptable. Converged protocols are a key ingredient to an SOI implementation.