Data center modernization depends, first and foremost, on addressing existing legacy applications and systems. Asset management has become a financial problem, and many initiatives aimed at gaining control of growing information technology costs are ineffective because application and system assets haven’t been properly tracked. This is readily addressed through better management approaches.
Modernization includes the process of achieving efficient use of all platforms, matching workload to the computer most suited to managing it. Implementing the new System z in combination with use of specialized, dedicated or hybrid systems as workloads dictate provides for an optimized solution. As Dot Alexander, senior analyst for WinterGreen Research and retired IBM executive, notes, “Many customers overlook the business and economic advantages of what modernization approaches can do to enhance the value of legacy assets while they pursue a ‘rip and replace’ approach that’s costly and less robust.”
With IBM’s announcement of a new-generation mainframe, the zEnterprise 196, information technology is moving to embrace a more tightly coupled use of System z with Linux/UNIX/Windows (LUW) servers and appliances. In this context, information technology initiatives are centered on a modernization theme. The modern information technology approach calls for a workload-centric, workload-optimized systems solution in which the mainframe is used for shared workload, and servers connected to the mainframe are used for processor-intensive offload of specific workloads.
Modernization requires gaining control of legacy issues. New workloads on System z should be able to take advantage of Service-Oriented Architecture (SOA) to accommodate a flexible response to changing conditions. Through modernization, legacy systems can be upgraded to take advantage of SOA and newer, faster, more flexible System z mainframes that efficiently use interconnected servers, offload engines, and enhanced software to manage workloads.
Concepts to Consider
Here are a few proven concepts you can apply as you pursue modernization:
- Improve mainframe utilization with shared workload. Many of the most advanced data centers have been able to concentrate on using the IBM System z more efficiently as they implement a modernization program. Shared workload, the hallmark of the System z, is more efficient than discrete servers or even virtualized servers in mixed workload environments. With shared workload, System z runs at 85 percent utilization with 1,500 images and another 3,000 applications, and provides a reliable and secure environment that’s far more efficient than what’s achievable on individual servers with five to 50 images each and 15 percent utilization. Shared workload and high utilization are essential when building a modern IT data center. High utilization can be achieved using offload engines and LUW servers when there’s processor-intensive workload running on a dedicated platform. This is what’s being called hybrid modern IT systems.
- Identify applications to keep or phase out. Often, the modernization process can lead to significant savings. As businesses implement modernization, they update older technology and consolidate or decommission some applications. The aim is to simplify the information technology implementation to eliminate redundancy, reduce maintenance, and lower costs. Evaluation of application portfolios needs to occur regularly to determine where information technology is delivering business value. Weighing the costs and benefits of each application can help when deciding which applications should be modernized.
- Remove the dead wood. Many old systems have lines of legacy code that aren’t used and old data files that sit unused; these all need to be discarded. Unused LUW servers that have been unplugged may still have legacy software licenses that need to be turned off.
- Conduct an ROI analysis. An ROI analysis is helpful. Figure 1 shows how an ROI calculator is useful for looking at whether a modernized System z can outperform servers or vice versa for running mixed workload with heavy database calls. When there’s highly virtualized, Web-based workload or shared workload with many different applications and database queries, the System z is far less expensive than LUW servers for any situation where more than 50 LUW servers are needed.
- Use dedicated servers to offload some mainframe processing. The use of dedicated LUW server processors for large processor-intensive workloads is justified. The LUW servers may run at 80 percent utilization and manage the workload more efficiently than the System z can because the server is running a single workload and is optimized for that workload. The same is true of offload engines.
- Pursue scale-out processing. Underutilized servers can be consolidated onto mainframe Integrated Facility for Linux (IFL) virtualized server images if they’re Linux, or a large blade if they’re Microsoft-based.
When analyzing the potential for shared workload and conducting your ROI analysis, you may consider MIPS usage, which is frequently viewed in terms of a four-hour rolling average. Shared workload calculations need to be done, looking at usage from various views (see Figure 2):
- Calculate the number of MIPS used at peak by strategic component applications and show what other workloads run “for free” in the non-peak whitespace within cost allocations.
- Calculate the number of MIPS per application component and allocate that component as a shared workload percentage of the total of all workloads. This is another way of looking at shared workload. Every workload must be accounted for. If a workload is migrated off the mainframe, then there need to be servers to host all those workloads and that capacity must be accounted for. When modernization ROI analysis occurs, all the workloads need to be considered, no matter how many there are and whether or not they run in whitespace.
Some Helpful Solutions
Applying the preceding concepts is a good place to start, but the list isn’t exhaustive or highly specific. As part of your modernization initiative, you may wish to consider some of the following products, which may be valuable for some situations and requirements:
- The IBM Optim software solution can be used to perform application retirement and data archiving and has significant capabilities to augment modernization.
- The Progress Software DataDirect Shadow product can outperform both the System z General Purpose Processor (GPP) and LUW servers for certain workloads. Shadow can optimize Web services invocations and database calls on the System z Integrated Information Processor (zIIP) offload engine, creating significant economies of scale in the context of those particular workloads.
- The ITP-Panorama system. ITP Panorama is one of several companies offering online/real-time documentation and information. Tooling provides a complete mapping of complex job streams, permitting modernization to proceed with elimination of unused lines of code. ITP-Panorama is unique in its ability to analyze and document components of complex, heterogeneous systems as an entire view, leaving no doubt as to what’s used code and what’s unused code and relationships of software and data components within the defined environment. The ITP-Panorama system is useful to modernization because it works in a matter of hours and is clear and easy to use. The tool scans source code and creates a repository of data concerning the programs and associated data files, stored as a value chain and data flow analysis representation in a Windows server-based Online Analytical Processing (OLAP) hypercube. The speed of scans allows for daily updating of the hypercube during modernization projects. This hypercube technology identifies all the paths and links within all code in use (and not in use) in a data center and provides insight that’s key to modernization efforts.
Modernization can take many forms (see sidebar), but fundamentally, it’s the process of evolving to more efficient applications without rewriting the applications. The process typically reduces the quantity of legacy code. Old, seldom or never used data can be compressed and archived; some old or unused code and applications can be retired. The process can also yield better workload management involving the use of modern systems on the new System z with well-planned integration of offload engines, servers acting as offload engines, and appliances as part of a hybrid data center that uses different platforms as workloads dictate.
Modernization facilitates operational efficiency. For data centers to decrease operating costs, modernization must occur. It isn’t wise to sit contentedly with legacy systems. As critical business data is scattered among application instances, consolidation needs to occur; it offers advantages in promoting consistency. Decommissioning unused applications offers cost advantages. In removing a system, application, database or hardware platform from service, resources become more efficient, lowering infrastructure costs and mitigating operational risks.