Live servers jumped more than fivefold from 6 million in 1996 to a stunning 32 million globally in 2007. They were mostly distributed servers. This sprawling flood of servers had dire effects. Blame the RISC-UNIX and x86/x64 cartel vendors for hyping and pushing these distributed servers to users. This last decade of wild proliferation in scale-out distributed computing drove the huge increases in IT operating costs detailed below.
These soaring distributed server units drove such IT management costs as software, staffing support, and data center up six-fold from $22 billion in 1996 to an unsustainable $120 billion in 2007—and they’re still climbing fast. IT electrical power/cooling costs also shot up from $10 billion in 1996 to $22 billion in 2007—also still growing fast. Combined, these IT system running costs were triple those of all global 2007 new server spending!
Supporting, managing, and administering such overly complex IT infrastructures of thousands of scaleout servers and their associated storage, networks and applications, and powering and cooling them, becomes astronomically costly. These high, annually recurring, ongoing support and power costs now far outweigh global customer expenditures on new server systems threefold.
The distributed server vendors sold 50 million x86/x64 servers in 10 years, plus all their associated software, networking gear, storage, etc. Yet 90 to 95 percent of this hardware/software capacity was completely wasted and never even used. Users couldn’t productively benefit from more than a tiny 5 to 10 percent part. Yet, IT operating costs vastly grew. An astonishingly profitable business for IT vendors, but $100 billion was wasted by x86/x64-UNIX distributed, server-addicted IT users!
These distributed system nightmares make IT infrastructure optimization the number-one job for most IT groups today. This is the latest step in a decade-long, continuing IT infrastructure improvement process. Four prior consolidation waves pre-dated the current distributed systems focus, and three other, concurrent waves are now often implemented alongside. These, in date order, were:
• Network Standardization & Consolidation
• Distributed Server Consolidation/Virtualization
• Data Center Consolidation
• Enterprise Storage Consolidation
• Large Systems/Server Consolidation
• Enterprise Application Instance Consolidation
• Distributed Servers Physical Consolidation to Server Farms.
Scale-out distributed servers are like the crack-cocaine of the IT systems business— highly addictive and extremely damaging. Once an enterprise gets hooked on this potent drug, it has proven extremely hard to wean the corporate addicts off! Despite the devastating consequences, they reached for their next x86/x64 and UNIX scale-out server fix for every new workload that arose.
Collectively, the distributed systems ecosystem vendors sold more than 50 million distributed x86/x64 and RISC UNIX volume servers in the past decade, plus associated storage, operating systems, middleware, and application software (the latter far more than server hardware). Yet, a large majority of the capacity of this vast amount of hardware/software sold was completely wasted because it could never be productively used by the customers.
The scale and depth of issues arising from today’s complex distributed infrastructures clearly require further IT wide-ranging IT infrastructure optimization and consolidation at most major enterprise IT users, and on a more modest scale at smaller firms.
The economic and financial reasons to quickly eliminate this massive waste of costly hardware and software, support-staffing effort, data center space, and power and cooling energy costs, combine to make a hugely compelling case for drastic consolidation and optimization. IT energy consumption absorbs two percent of global electricity production so an enterprise’s IT energy reduction through drastic IT consolidation is a major and vital part of corporate green initiatives to reduce enterprise carbon footprints and also help limit global warming in many industries.
With their myriad serious issues, IT user infrastructures afflicted by scale-out distributed computing addiction/disease now imperatively need radical, rapid consolidation and optimization at thousands of enterprises globally. The most compelling alternative is extreme virtualization and mass distributed server workload consolidation using Linux on z/VM (hypervisor) on IBM’s System z platform. We will discuss the striking benefits of System z10 mass consolidation in a later column.