Latest Entries

The 10 commandments of TCP/IP performance are a distillation of hard-won experience. Monitoring and tuning TCP networks on the mainframe is complex for the basic reason that each network is a mixture of many applications and pieces of hardware. Like an onion, each connection contains layers of protocols and subprotocols that must be decoded to make sense of the traffic patterns. Making sense of it all is the first step to tuning and improving performance…

Read Full Article →

SOA (services-oriented architecture) is an established development methodology today because it equips companies with agility as they respond to rapidly changing business conditions. It also offers a way to reuse software in a multiplicity of present and future system and application contexts that conserve IT costs. SOA’s modular software reusability is facilitated because software components like Web services are decoupled from tightly integrated legacy systems. In theory, this reduces the risk of software changes impacting other areas of software that they interface with. But as software becomes more and more decoupled, is there a corresponding risk that overall change management also gets decoupled?…

Read Full Article →

Common wisdom in the industry holds that mainframe professionals comprise a demographic with an average age of 56. Ageism, therefore, probably plays a role in the perception of mainframers by senior managers, who increasingly make decisions about the future of corporate IT. As “isms” go, ageism is actually a neutral term. It refers to a predisposition about age that can be either positive or negative. It’s sometimes “anti-age” and sometimes “pro-age.” Let me explain. …

Read Full Article →

By the time you read this, the late 2009 version of the annual IBM Systems Technology Group (STG) analyst briefing will be a distant memory. Much of the messaging at this briefing was similar to years past. Cost containment? Been there. Cloud computing? Not too far from the hype about Software as a Service (SaaS) from five to 10 years ago. But there was one new theme that dominated the discussion: mainframe strategies. …

Read Full Article →

Suppose we asked, “Will the real mainframe virtual tape functionality please stand up?” Four candidates—four unique, yet complementary, solutions—would stand up. All four could legitimately claim to perform a tape virtualization function in a z/OS environment. Although a number of companies have products in the virtual tape space, IBM will serve as an example here, as it covers all four solutions. I recently took an analyst briefing from IBM on the new ProtecTIER solution (another “face” to tape virtualization), so I thought it was a perfect time to examine all the faces of tape virtualization.

The original use of virtual tape was (and is) to more fully utilize the capacity of physical tape cartridges. Historically, the process of writing data sets directly to tape cartridges has been very inefficient (say one application), which resulted in much of the capacity of a directly written tape cartridge being unused. With the average capacity of tape cartridges rising rapidly and no end of this trend in sight, writing directly to tape continues to be a worse idea each year.

The first use of virtual tape—to collect and concatenate data on disk to form large, virtual volumes for better space utilization of tape cartridges (i.e., tape stacking)—has proved to be an acceptable, well-regarded alternative. The result is that a physical tape library and the number of tape cartridges can be sized as close to optimum efficiency as possible.

However, the use of disk cache as a front-end to tape led to a second tape virtualization solution. Using intelligent management software on a server (what IBM called a Virtual Tape Server [VTS]), this second virtualization solution broke the connection between logical and physical tape drives, allowing storage of “virtual tape volumes” either on disk or tape. Physical tape drives are expensive compared to virtual tape drives. So users of the second solution can now “manage to need” (dynamically determine how many “tape drives,” physical or virtual, are needed at any one time to keep all job streams working most efficiently) rather than “manage to cost” (determine how to use the capacity of all available physical tape drives most efficiently to postpone the purchase of a new drive for as long as possible).

The IBM Virtualization Engine TS7700 family is a follow-up to VTS. Whereas IBM’s VTS products are peer-to-peer, the new family uses a grid architecture to considerably scale both performance and capacity.

A third use of virtual tape emerged as a z/OS software implementation useful for certain niche applications. IBM Virtual Tape Facility (VTF) for Mainframe (VTFM) uses mainframe-attached 3390 disk for virtual tape storage, thus providing time-sensitive tape applications with consistent high performance and virtual tape replication completely consistent with all replicated disk data, including catalogs, control files, and indexes.  

IBM’s VTFM feature set, scalability, and multiple implementation options make it a very useful, general purpose virtual tape solution for clients with smaller tape libraries or for clients with unique tape processing needs. One such feature, unique to VTFM, is Parallel Access Tape (PAT), which allows multiple applications concurrent read access to data on its virtual tapes.

The fourth use of virtual tape is the data deduplication capability that’s derived from open systems “virtual tape library” deduplication. Data deduplication takes advantage of the fact that much of the data written to tape is for backup data protection and that most of the data written over time is the same, and eliminates this duplicate data.

Compared to the second and third approaches, data deduplication saves additional disk space. It also means a shop can keep older data on disk as well as current data, offering the ability to partially or totally eliminate the need to go to physical tape in order to find the correct data to restore. This fourth use of virtual tape is exemplified in the new IBM System Storage TS7680 ProtecTIER Deduplication Gateway for System z.

Users shouldn’t think of these four solutions as “either/or” choices. Rather, they form a spectrum from physically all tape to some disk or lots of disk; the use of physical tape technology is mandatory for the first solution, but not the other three, and all allow disk caching. Thus, users can employ each solution for a different segment, or tier, of their tape library, depending on such factors as performance, duration of data retention, number of copies maintained, access patterns, cost, manageability, and local and remote recovery/restart needs. All four solutions could stand up as tape virtualization solutions.

Read Full Article →

Web services are an increasingly common design paradigm, and where these applications are to be provided is the first design decision. IBM’s CICS Transaction Server may not be an obvious choice, but CICS is a good environment for hosting some Web service applications. CICS isn’t a general-purpose Web server; it’s a 40-year-old transaction processor. Its utility for Web services boils down to secure, scalable access to existing applications and data. The Web services of interest are those based on SOAP/XML, and CICS can act as a Web service provider for these services…

Read Full Article →

Traditional mainframe roles are undergoing a metamorphosis that transforms the emphasis from product-centric specialists to a multi-role generalist. With budget cutting and the mandate to “do more with less,” this transformation extends to Database Administrators (DBAs) tasked with multiple responsibilities. Often, DBAs must shift from generalist to specialist and back, and always in an increasingly dynamic, complex environment…

Read Full Article →

An Overview of CICS Channels and Containers

Ever since command-level programming was introduced in CICS in 1975, an in-memory process called the communications area (COMMAREA) has been used to pass data from one application to another. From its inception, the size of the COMMAREA was held in a half-word data field— the EIBCALEN—which limited the size of the COMMAREA to 32KB. Although this was a problem, since most application  data was 3270-based and not, therefore, particularly resource hungry, the 32KB limit was sufficient to manage data propagation between applications. However, with the advent of Web-based applications, today’s CICS applications are required to process larger quantities of structured parameter data in both XML and non-XML formats, meaning the constraints imposed on the COMMAREA size by the 32KB limit are too restrictive. Furthermore, if an XML document has to be exchanged, the parameter data contained in it is quite different from the data format historically known to CICS programmers…

Read Full Article →