DataDirect Quantifies Cost, Value of zIIP-based SOA
DataDirect's new zIIP TCO calculator shows just how useful IBM's zIIP engine has been to mainframe ISVs.
Since its introduction more than three years ago, IBM Corp.'s zSeries Integrated Information Processor (zIIP) has been a boon to both customers and mainframe independent software vendors (ISV). ISVs, in particular, have embraced zIIP as a way to help customers reduce their software licensing costs -- chiefly with respect to how much they pay Big Blue for Big Iron capacity (see http://esj.com/articles/2007/02/20/cas-ziipy-masterstroke.aspx).
This month, mainframe ISV DataDirect Technologies -- along with consultancy WinterGreen Research -- unveiled a new Web-based total cost of ownership (TCO) calculator that it says can demonstrate the TCO benefits of mainframe service-enablement by using the zIIP engine in combination with DataDirect's Shadow integration middleware suite.
"It's a Web-based [tool] that computes the daily and annual savings in dollars and MIPS capacity that [customers] can realize by exploiting zIIP in combination with Shadow," comments Jeff Overton, senior product manager for Shadow with DataDirect.
The idea, Overton says, is to shift appropriate SOA-related workloads from the mainframe's pricey General Purpose Processor (GPP) over to zIIP; in Shadow's case, he claims, DataDirect has succeeded in moving almost all SOA-related processing over to zIIP, including SOAP requests and as event publishing and data processing workloads. According to Overton, DataDirect has been able to shift 99 percent of such processing over to zIIP.
There's a clear sense in which DataDirect's new zIIP TCO calculator illustrates just how useful zIIP has been to mainframe ISVs -- just as it introduces still another intriguing application of zIIP in a non-traditional -- or, at least, unconventional -- context. In this respect, it's similar to how other vendors (such as mainframe software giants BMC Software Corp. and CA Inc.) have been treating zIIP since its introduction.
Consider CA, which -- about a year after zIIP's official debut -- touted what it billed as an unconventional (but nonetheless sanctioned) take on zIIP's specialty engine benefit: the ability to run quasi-extra-data processing workloads. CA trumpeted several clever takes on "zIIP-i-fication," such as offloading tasks such as network packet analysis or data encryption from comparatively expensive mainframe general processors (GP) onto zIIP.
CA officials acknowledged that (at first glance) Big Blue's zIIP terms did seem too good to be true. "What IBM has outlined is a very technical list of requirements of the work you want to run on zIIP, and if you meet those requirements it really doesn't matter what the work is, as long as it's running in the right modes and states," said Vince Re, senior vice-president and chief architect with CA, in an interview at the time. "A lot of customers were confused by this, [and] IBM's own announcement was a DB2 announcement, but that's not really the case. It's just the case that DB2 met all of those requirements. It really isn't a DB2 or a data-serving thing, it could be anything."
The point, Re asserted, was that CA's unconventional take on zIIP nonetheless strictly adheres to the letter of Big Blue's zIIP licensing terms.
Data Direct's Overton isn't comfortable with the notion of zIIP as a giveaway. "That's not how I see it, and I really don't think that's what [IBM is] doing," he says. At the same time, he concedes, there's a sense in which Big Blue has been charitable with the terms of zIIP, which -- because vendors such as BMC, CA, or DataDirect hew to IBM's published zIIP guidelines -- has proven to be a surprisingly flexible engine.
"For ISVs, [IBM has] very specific requirements on its zIIP [Web] site. As long as you stay within those [requirements], they're fine with whatever you do," he comments. "Some of our competitors are using [zIIP] to run maybe 25 or 30 percent of their [data integration] processing. That wasn't good enough for us. We saw that with zIIP, if we did a thorough job, we could get as much as 99 percent of Shadow to run [in a zIIP context], so that's what we took our time doing. If you really study [your application], you can figure out just what you need to do to get it to run [in a zIIP context]."
Overton concedes that being able to shift SOA and related data processing workloads from a GPP to a zIIP context doesn't necessarily affect what customers must pay DataDirect in order to license Shadow, but it does make the cost of service-enablement -- which, he insists, offers TCO benefits of its own -- considerably more palatable.
"For someone who's looking to achieve [SOA and data integration] efficiencies with what, in many cases, are substantially reduced budgets, the ability to exploit zIIP like we do with Shadow can make the difference between being able to do it -- being able to justify it both in terms of upfront cost and potential return on investment -- or not being able to do it," he points out.