Mainframes and Wintel: Still in Search of the Grail
From Byzantine stories of a magic stone, to the Celt’s magic cauldron, the Arthurian chalice and Dan Brown’s missing link in “The Da Vinci Code,” there is little agreement on what the grail is, much less where to find it.IT emulates the grail quest, except IT’s quest is for the “right” technology. How often has IT uttered the plea, “if we migrate to ‘x’ technology, everything will be better?” Yet, IT’s attainment of the grail continually disappoints. Past decades provide ample proof: ENIAC, 360/370 Assembler, COBOL, VSAM, relational, UNIX, “C”, client-server, O-O, e-business and now Java and SOA. IT quests for a grail it can never fully realize because the technology changes faster than firms can reconcile it.
WFS: Can Wintel platforms replace mainframes?PM: Wintel servers are now allegedly powerful enough to run the mighty workloads that once were the exclusive dominion of big iron and OS/VS COBOL. Firms move COBOL source to Wintel and emulate the missing parts such as CICS, JCL and DB2 with COBOL compilers available from Micro Focus, Fujitsu and other third-party vendors.
Forrester research shows that smaller firms, e.g., those in the 75 to 200 MIPS range all reported cost reductions that far exceeded their reported mainframe costs, primarily on eliminating third-party software costs. However, 200 or 300 MIPS is far from the upper limit. At Forrester, we’re aware of firms in the 800 to 1,000-plus MIPS range that have moved to Wintel and enjoyed similar results, some with reported savings ranging from $250K to $400K to $1 million or more. Most of the interviewees were pushing the MIPS limits of their existing mainframe environments, and post-move, report faster processing. So, the answer to “can you move?” could well be, “yes.”
WFS: Where are the holes in the story?PM: Despite the good news, Forrester’s research unearthed complaints of operational immaturity that have dampened the enthusiasm of some, such as a financial services firm that moved a 450 MIPS workload to Wintel, and a logistics firm wanting to quickly exit a Honeywell-Bull mainframe. Both lament the lack of a number of features:
Database tuning. Because SQL Server behaves differently than DB2, a retuning effort is necessary.
Lack of options for batch job scheduling. Scripting was simply not enough.
Weak batch processing support, especially for complex job step interdependencies.
Weak support for batch invoice printing. They required customized invoices for each customer.
More complex configuration with less available automation. One interviewee noted that adding space to the DBMS is problematic, saying “on the mainframe, you just throw the space at DB2 and it figures out where to use it, versus manually allocating it across dozens of Wintel servers.”
The odd take-away from these complaints is that the market seems to have learned little from operational maturity problems that plagued the mainframe-to-Unix migration craze of the late 1990s. Many of the same operational gaps that existed then remain unaddressed 10 years later. Vendors in the Wintel camp: take note of that.
WFS: Should Wintel platforms replace mainframes?PM: As general rules-of-thumb, 1,000 MIPS and up pushes the known limits of mainframe-to-Wintel moves, which is not to say that 1,000 is a known physical limitation. Under 200 MIPS, processing requirements seemed to be simple enough in those firms interviewed, to suggest a high likelihood of significant cost reduction. Between 300 and 1,000 MIPS, while success depends on the existing level of application-processing complexity, especially batch processing, there is no one-size-fits-all answer.
At Forrester, we suggest that firms look beyond their tactical grail searches for the “right technology” and move toward a more strategic review of applications, based on an application scoring mechanism so that a clearly defined application strategy, not tactical technology choices, rule IT’s future actions.
Bottom line? End the quest by planning an application strategy, starving commodity applications to feed core competency applications.
Phil Murphy, a principal analyst in Forrester Research’s IT Leadership research group, focuses on the technology and processes enabling firms to analyze and quantify their application portfolios. He has written several research articles concerning strategic decision-making about which applications to retire, redevelop, modernize or outsource.
