MAINFRAME WHITEPAPER

WHY ARE MY MAINFRAME COSTS CONTINUING TO INCREASE?

It is no secret that IBM Mainframe costs have been increasing.   What’s worse, 3rd party mainframe software costs from vendors such as Broadcom have been increasing even more.

And there’s no end in sight.   IBM has a monopoly on the hardware.  The mainframe software used is concentrated into a handful of vendors.   And it’s very difficult to switch, leaving most mainframe users at the mercy of these vendors.  It’s no wonder that most will continue to be challenged in managing their mainframe costs.

But are these the only reasons costs are increasing?  No.

  • As your business logic and use cases change (often becoming more, not less, complex) transaction counts and costs can increase.
  • New versions and releases of third-party software that have additional security and other features added can result in transaction counts and costs increasing.
  • Like all compute platforms, It’s not uncommon for MIPs to grow on an annual basis even though revenue doesn’t increase because of these additional transactions.

The bottom line is, if your business is growing, your transactions counts are growing.   And therefore, so are your costs.

Is this a bad thing?   Possibly.   But so long as the following two data points are true, maybe not:

  1. The mainframe is the most cost-effective platform to run your workload
  2. Your increase in total MF cost is less than the corresponding additional revenue being brought in

FACING THE FACTS

Mainframes will be around for at least the next 10 years, likely longer.  How long?

Until the price/performance of Server/Cloud surpasses that of the mainframe, for workloads of size and complexity.

For organizations that use the mainframe for mission critical applications, such as banking, insurance, supply chain, manufacturing, etc., the rather inconvenient truth is that byte for byte, the mainframe is still, in most cases, the fastest, most secure, and most cost-efficient way to run those applications.

Those preaching a massive conversion to newer technologies running on the Cloud also know this but are betting that the current narrative of “mainframe bad, Server/Cloud good” is strong enough to convince you to not perform an objective and detailed cost-benefit analysis.

It takes a pile of money and a long time to convert.

According to ChatGPT it costs on average between $1 and $5 per line of code to convert to Server/Cloud, and we’ve heard of costs up to $20/line of code from some consulting firms.

It’s also worth mentioning the obvious: Amazon, AWS and Google Cloud get paid when clients consume their cloud services. That’s why they often recommend lift and shift to quickly get their meters running so they can charge for consumption of their infrastructure. These three Cloud Services Providers are less in the application modernization and cloud migration business as much as they’re in the “move your workload over to our infrastructure business”.  Doing the math: large applications out there START at 10 million lines of code and can reach into the 100’s of millions of lines, so it’s no wonder that conversions are not for the faint of heart.  Not to mention that these conversions generally take years to complete.  We have not even broached the topic of increases in day 2 run costs.

This is why most CxOs are very hesitant to green light such projects.  That’s a lot of time and money spent just to try and get what you’ve already got.

While all this sounds like doom and gloom, it is not.

Faced with the fact that mainframes are not going away yet, we turn our attention to what can be done to manage mainframe-related costs and get much better platform value.

Surprisingly, there are quite a few things that can be done.

EFFECTIVE MAINFRAME COST MANAGEMENT STRATEGIES

Surprisingly though,  the truth is that with the right team, you should absolutely be able to take out between 10% and 30%.   

And a lot of it without touching any hardware or coding any software.

We will touch on a broad variety of approaches.  But the focus and key takeaway from this paper is that there are a whole lot of different ways to manage and reduce your mainframe costs simply by having an in depth understanding of:

  1. Your environment
  2. Your contract terms and conditions
  3. What IBM (and other vendors) will and will not do
  4. Your ability to negotiate outside the box

Your strategic sourcing group and your mainframe team may tell you that they’ve already done everything possible including what we discuss here.   Maybe they have.  Maybe they believe they have.  But unless your costs are dropping by 10-30%, they have not.

Here are the four main categories of mainframe cost and their approximate % of the total mainframe related cost:

  1. IBM hardware costs (20-30%)
  2. IBM software costs (15-30%)
  3. 3rd Party software costs (15-20%)
  4. People costs (30-40%)

REDUCING IBM HARDWARE COSTS

This is a difficult category of mainframe spend to manage because IBM has a monopoly on most of the hardware.   Which they take full advantage of.  In fact, they have a wonderful (for them) strategy of planned obsolescence where you are forced to upgrade at regular intervals.

Without in-depth knowledge of how IBM pricing works, the only saving grace for many companies is that the purchase of hardware is a capital expense which is preferrable to OpEx.

For those in the know, there are a few strategies one can use to try and reduce mainframe hardware costs:

  1. It will come as no surprise that the largest buyers of MIPs get the lowest price. What may be a surprise is that IBM sometimes prices Development, Test and Disaster Recovery MIPs less than Production MIPs.
  2. In addition, MIPs that are restricted (for example; MIPS dedicated for non-revenue generating purposes like queries of account balances) can be priced less than Production MIPs.
  • Although MIPs can’t be moved from CPU to CPU it’s possible to purchase something called Flex MIPs which can be moved between processors for predetermined times, resulting in far more cost-effective usage, and less leakage, of each MIP.
  1. There can also be cost relief through Rolling New Technology leases.
  2. Refactoring your environment to take advantage of virtualization and containerization, which often results in less resource usage, and therefore lower costs.

The more you share with IBM about your UNIQUE mainframe requirements, the more IBM will consider targeted  reductions in MIPs pricing.  Naturally none of this comes free.  You will need to motivate Big Blue to give you these concessions.

How do you motivate IBM to do the above without increasing your own costs?  Ay there’s the rub.  We address this later.

One other cost savings approach that does not usually immediately come to mind, is to buy the whole mainframe estate as a service.

This entails handing off or ‘outsourcing’ the entire estate to a third party, in effect SAASifying the hardware cost – moving it from a one-time spend CapEx to reoccurring spend OpEx.

This can significantly stabilize cashflows  and insert a big chunk of cash into the balance sheet when you sell your estate to your new service provider.  Any CFO reading this is smiling.   We won’t sugar coat it though.  Doing this is a lot of work.  With its own pros and cons that we save for another day.

REDUCING IBM SOFTWARE COSTS

IBM has, over time, done a masterful job of convoluting the terms and conditions of its software, to the point where it is extremely difficult to follow all the moving parts.   All of which benefits them, as many simply give up, sign the boilerplate terms, and pay the price.

But if you know what you’re doing and have deep IBM knowledge available to you, (which does not include golfing with your IBM rep),  there are ways to quickly and significantly r Capeduce cost here.

Successfully executing the below strategies can generate massive savings on a year over year basis.   Large estates can easily spend over $100m per year, which is $300m over the typical 3-year deal life cycle.   Taking out 10%-30% will save you a cool $30m to $90m over the term.

Some of the strategies we’ve used to successfully take out 10-30% of IBM S/W related costs:

  1. CBA Enhancements:
    • Expand what CBA can be used for (ideally whatever you want)
    • Rollover / End of term options
    • Credits for turning down software
  2. SPA/ACP Enhancements:
    • Useability against audit findings
    • Maximize s/w allowed in the SPA/ACP
    • Maximize SPA/ACP discount
    • List price price-hold
  3. MLC Enhancements:
    • Price hold
    • Maximize MLC discount
    • Perpetual (VUE) vs SAAS pricing
  4. ELA Enhancements:
    • Audit hold
    • Value of Early Renewal
    • Rollover / End of term options
    • Support pricing
  5. Using zIIP Specialty Processors:
    • to support “New & Relevant Workloads” such as Java application programs, z/OS Container Extensions (zCX), Watson Machine Learning for z/OS, Systems Recovery Boost, and selected DB2 processing (e.g. XML distributed queries and some utilities),
  6. Consolidating Workloads based on the quirks of your s/w license terms:
    • Vendor contract terms are designed to constrict you in such a way that you will need to spend more money over time
    • Being able to manage to these terms can save you from spending unexpected additional amounts
  7. Running Linux on the Mainframe (zLinux)
  8. Reviewing existing S/W on an individual basis for opportunities:
    • Upgrading / consolidating / downgrading / changing
    • Each large software deal is an island unto itself with its own set of opportunities to optimize
    • Time needs to be spent here, ideally with specialists who understand each application
  9. Leveraging the value of IBM Cloud:
  10. Proactive Audit Management:
    • Audits are a software vendor’s secret weapon. Compliance to their licensing terms is often non-existent and they know it.
    • One of the cheapest, fastest and easiest ways for a vendor to generate additional revenue
    • Managing this area properly can save you a lot of money, and it’s unbudgeted money so even more valuable to upstairs
  11. Application Tuning:
    • We put this one last and hesitate to do so because this is the first thing that most large consulting firms will recommend, because it’s a cash cow
    • However, effectively done, which means targeting only those applications that are high volume and have a proper cost-benefit analysis, can generate cost reduction

The ability to properly execute the above strategies requires a deep level of knowledge of IBM products, tactics and strategies.

This level and depth of knowledge is rarely found within the mainframe or procurement departments of most companies.

REDUCING 3RD PARTY SOFTWARE COSTS

This is one of the most challenging and contentious areas right now for many organizations that are mainframe heavy.

IBM mainframe 3rd party vendors have consolidated recently, and those companies are seriously raising their prices, with a ‘take-it-or-leave-it attitude.  It is not unheard of to see 50-100% cost increases.

As a primer, there are a few companies that have over the decades specialized in software for the mainframe.   Most of the time their only competitor is IBM.

Great you say,  there’s more than just one provider so I’ll just RFP my way to the bank.  Easier said than done.

While indeed you can switch to or from IBM for some categories of software, the challenge is the amount of time it takes to do that migration.   This type of software is sticky.  And as the CFO will ask, what value does it bring?   If your answer is not “our cost savings more than cover the migration costs”, you might as well pack it in.

Some Options:

  1. Buying Your Business:
    • It is sometimes the case that the vendor that does not have your business is willing to buy (give you concessions so you move to them) your business. With the right offer, IBM and others have been known to go as far as covering transition costs.  The trick is coming up with the right offer.
  2. Switching Licensing Models:
    • Companies go through cycles of offering perpetual versus subscription-based pricing. Sometimes they have a strong preference for one over the other and will offer concessions for you to switch.
  3. Managing Maintenance:
    • An oft overlooked area of cost is the insidious “maintenance” fee that comes with almost all software. It’s there whether you have a perpetual or a subscription-based agreement.   And it’s mostly gravy for the vendor.
    • Having an in-depth understanding of the terms of the maintenance fee can allow you to propose alternatives and/or restructure how you use the software to optimize.
  4. The Volume Play:
    • If for some lucky reason you have an option to increase your business with the vendor, you can likely negotiate a cost reduction on existing spend.
  5. Terms And Conditions Matter:
    • And finally, an area that is seriously overlooked by all but the best procurement organizations:
    • Ensure that you have some great terms and conditions in your agreements that allow you to hold prices at renewal, terminate for convenience, M&A optimization, referral leverage, free or discounted transition assistance, and many more. These are the types of terms that may not matter when buying, but matter more than any upfront discount, under certain future scenarios.
  6. Asset Management:
    • No discussion of mature software management is complete without talking about the value of strong asset management.
    • Vendors love organizations with no asset management.
    • Firstly, they know that usage is not tracked, will grow, and will generate huge proceeds if an audit is performed. Knowing exactly how many units you have, who has them, where they are, how they are being used can allow you to manage usage (and cost) tightly.

REDUCING PEOPLE COSTS

CIOs often state they are unwilling to reduce headcount and people costs due to risk of performance impact. But given people costs make up to 40-50% of the total costs of mainframe computing, leaving this area untouched means you’re leaving up to half of your potential savings on the table.

By implementing a disciplined strategy of only allowing a single brand of storage, a  single version of each software product (not allowing multiple versions of software), a single database, and other components, and doing other things like running Linux on zIFLs instead of other distributed platforms,  then you’ll need fewer people to manage these different products.

Unique to the mainframe environment,  you can often double the MIPs installed without adding additional support headcount. By comparison, if you double the size of your distributed server or server/cloud environment you might have to add up to 25% more headcount to support it.

Also the discipline of strict adherence to  standards  can help reduce headcount by 10 percent or more of the people who run the mainframe, people who maintain the mainframe applications, and people who maintain the mainframe databases.

Interestingly, the one way  to get rid of the need for ALL your mainframe folks, is to do what we talked about earlier:  Outsource the function.   By doing so you will be able to leverage the outsourcing vendor’s pool of talent, which is far larger and more stable than your own and take all their direct cost off your books.

THE SECRET TO EVEN MORE MAINFRAME COST REDUCTION

If you are still with us, we are first and foremost thankful, followed by impressed.  We assume anyone still reading really and truly has a mainframe spend problem, and really and truly wants to do something about it.  So here is your persistence award – an additional way to reduce your mainframe spend:

Reduce spend in non-mainframe areas.

Think about it.  Cost reduction can come from anywhere.

Let’s assume for a moment that you absolutely need to reduce your mainframe cost by $100m in run rate OpEx.    At the end of the day, does it really matter if you take out $100m in mainframe, or take it out somewhere else?  Or

What if you took it out of your Network spend (which is a complete gold mine and a topic for another day).  Or your Application Development spend?  Or your Professional Services spend?  Or your Cloud spend.  Or your non-IBM hardware/software/services spend?  Or your facilities spend?

Taking cost out of your overall vendor spend is just as valuable and just as important.

And just like in the mainframe space, there are a ton of tricks and tools needed to win big in other areas of spend.

Bob Hoey and Raj Kapoor

WhiteStone Group  Inc. |  Spend Less.  Do More. | www.whitestone.group

The WhiteStone Group specializes in Mainframe and Vendor Spend Optimization and has taken out over $1.3B in hard costs for their clients.

Bob Hoey is a 30-year IBM veteran and the former IBM Global General Manager of Sales for IBM’s Systems & Technology Group, as well as the former IBM General Manager for the North American Financial Sector.  He knows more about IBM pricing than he would like.  He can be reached at bob.hoey@whitestone.group.

Raj Kapoor is the co-founder of The WhiteStone Group and has spent the last 25 years advising and executing on rapid cost takeout, and rapid technology transformation. He can be reached at raj@whitestone.group

Back To Top