Google (GOOG) has to be one of the most audacious companies in recent history. They took the advertising industry by storm, and as I wrote recently could even be accused of fomenting war with enterprise IT. Whether you like Google or not, you cannot help but be impressed with their audacious leap from search advertising and mysterious algorithms into the even more mysterious world of cloud computing.
Cloud computing has become the latest obsession in IT. And regardless of the paucity of revenues relative to headlines, it has stormed into the collective tech imagination like a Texas flood. As virtualization decouples applications from hardware, cloud computing promises to decouple IT services from hardware; and that threat alone is enough to send tech vendors scampering with slides, announcements and other forms of marketware.
Yet underneath the billowing vapor there are some interesting rumors, announcements and events. Google is rumored to be introducing a router; and Cisco (CSCO) is reported to be working on a blade server. I think there is a common thread between all of these developments, and cloud computing may be a mere mask for a core set of challenges facing enterprise IT and threatening the most well-intentioned and thoughtful initiatives. For the moment let’s call it “Peak IT.”
Peak IT occurs in an organization when investments in automation and productivity tools are crowded out by rising operations and management expenses. The result is a downward spiral of increasing complexity and expense and shrinking productivity as expenses continue to increase and take a growing share of budgets.
While cloud computing has been all the rage for predictions and billowing potentials perhaps this enthusiasm is more of a symptom of an ignoble shift in IT business realities than it is of a new tech cycle of optimism. The key driver behind the cloud dream may likely be the increasing realization that IT (network and server) per unit management costs are escalating and many CFOs and CIOs are becoming entranced by the thought of warehouse IT.
Cloud Fantasy amplified by Painful Reality
The cloud meme addresses a growing need for IT to get back on track from the standpoint of expenses and productivity or be a casualty of the “Does IT Matter” school of commoditization. Whether cloud is achievable or real today in many of its forms or just a sky-high fantasy doesn’t matter because IT at many enterprises is at a (peak IT) crossroads between rising per unit (servers, IP addresses, people) costs and declining influence over business productivity.
Check out slide 4 in this presentation on blade servers accessible through the Microsoft (MSFT) website. Based on an IDC report in 2006, it shows server management and admin costs increasing dramatically over time and ultimately dwarfing actual hardware costs (which stay pretty level). I would have shown the slide except for the “HP Restricted” disclaimer.
Yet here is another slide thanks to IDC that doesn’t show the time trend line but rather a slice in time. Note: manual labor and training counts for over 70% of server costs. That makes the case for automation and even outsourcing (or “cloud sourcing”) extremely compelling. And according to the research cited (slide 4 link) in the previous paragraph these costs are still increasing relative to hardware.
click to enlarge
This report is also consistent with last fall's Computerworld survey showing larger networks incurring higher (per IP address) costs for IP address management, or the manual labor involved with simply keeping up with what is attached to the network.
Networks and data centers get more complex and costly as they grow, and the effect is cumulative and exponential. More manual tasks, more committees and more errors are the natural result of diseconomies of scale as IT infrastructures outgrow the tools and processes put in place when days were simpler and systems were static.
Indeed, peak IT may be the number one competitor for every major hardware and software vendor, from Microsoft to Cisco and F5 Networks. As enterprises dedicate more manual labor to the maintenance of IT infrastructure they may have less to spend on new gear.
IP address management is a perfect micro example of the “peak IT” threat where costs grow exponentially and undo the economics of new “selective automation” initiatives like early virtualization and cloud computing projects. Systems are automated, yet processes and tools are still decades behind; costs are pushed into teams where the processes are still manual rather than reaching the balance sheet and fulfilling the vendor’s promises.
Journey to the Center of the Network
Jules Verne’s classic 1864 novel tells the story of a professor who takes his nephew further back in time as they get closer to the earth’s center. A CIO might have a similar real-world experience as he (or she) peels back the onion in the network and looks at the process checklists surrounding simple tasks like adding a printer (six steps involving three departments in many cases) or finding a range of available IP addresses from a wrinkled pile of spreadsheets complete with out-of-date comments and notations.
Let’s face it, most CIOs have limited real-time visibility into what is connected to the network and might have to ask multiple well-paid pros to engage in a project to merely collect that information; those reports will also be obsolete faster than ever because of new initiatives like virtualization and a new and even more amorphous cloud of swirling, billowing IT services being hyped beyond network reality.
Many of them are more knowledgeable about software and systems, hence the smart investments in application delivery and virtualization and a rising interest in cloud. This “selective automation” will only achieve small, incremental gains in one area (systems and server management and power savings) at the expense of another (network management operational expenses).
That is the base irony as the IT world gets caught up in yet another vision (or hype cycle) that stretches our imagination… and our capabilities. While servers can be created and moved by mouse click thanks to virtualization, network teams are still manually tracking IP addresses from spreadsheets, akin to the dying breed of enterprise CFOs who initially resisted the move to relational databases as their competitors started making real-time supply chain decisions.
As enterprises experience higher rates of change, their tools must evolve or their analytics and controls will become obsolete even faster. This is the essence of the CIO Shell Game of moving costs, escalating velocities of change and obsolete core network service tools and procedures. As costs move CIOs are further isolated from the current realities of their network infrastructure.
Cisco and Infoblox Embrace a New Vision
That is perhaps why more than 1100 IT professionals registered for the Infrastructure 2.0 event at the San Jose Fairmont, on very short notice. IT pros are hungry for a way out of the “peak IT” phenomenon plagued by rising management and configuration costs, even as we dream about IT warehouse clouds.
For more background you can view John Furrier’s (less than 2 minute) post-event interview with Cisco's Doug Gourlay or read a recent Gourlay blog on the challenges that cloud computing will pose to the network.
Infrastructure 2.0 is the necessary evolution of the network in the face of rising management expenses and higher velocities of change. It is every bit as significant to IT today as TCP/IP was during its early adoption. We received the following in an unsolicited email from marketing guru Peter Raulerson soon after the event:
I was watching something like the first Interop, when Dan Lynch brought together a couple dozen guys to plug their network products together to show that they really did interoperate, and 300 people showed up to see it. (Dan was a client of mine at SRI and I've known him since 1976).
Putting Infrastructure 2.0 within a Broader Context
The peak IT phenomenon probably isn’t that much different than the effects seen in other aspects of business and government where a complacent status quo reduces investments in innovation and/or automation for short term business goals. As network infrastructure grows, this triggers rising operational expenses (as IT becomes more operationally complex and experiences a rising tide of manual and procedural tasks that come with growth). A short term perceived “windfall” then sets in motion a precipitous drop in productivity and innovation as expenses increase further and productivity-related investments are crowded out by the commoditization of IT.
Peak IT Results from Lack of Innovation amid Rising Expenses
This productivity drop then puts increasing pressures on CIO budgets and sets up the IT team for increasingly impossible missions and rising network/application availability risks. CEOs and CFOs wonder what the CIO is doing with the budgets. Because as the peak is passed net IT investment actually becomes disinvestment, despite rising budgets. Yet every quarter past the peak robs the CIO of more budget for productivity.
That could explain why the cloud computing hype took off so fast in 2008 without any discussions of basic requirements (including the network and security, etc.); the pressures on the status quo to escape from the throes of the CIO shell game and post-peak IT maintenance drains were building in boardrooms. Consider this comment from InformationWeek (Bob Evans, January 2009):
In December 2008, McKinsey released a report called "IT's Unmet Potential" that calls out the 80% sinkhole in the 80/20 equation as a huge factor in keeping that potential from becoming reality. The McKinsey report says many CEOs are eager for the CIOs to be more assertive as "thought leaders" helping to enhance the company's competitiveness, but too often the CIOs can't take advantage of that opportunity because they are consumed by maintenance battles.
That’s why I think recent moves by Google, Amazon (AMZN), Cisco, VMware (VMW) and Microsoft are so interesting. Maintenance and manual processes are undermining ill-equipped CIOs and each vendor is using its competitive strengths to embrace virtualization and eventually cloud in order to break the cycle and benefit from a new era of automation: Infrastructure 2.0.
These companies and dynamics pull core network service automation vendors like Infoblox (my employer) into the vortex because reducing the existing expenses of the current 1.0 infrastructure is a critical enabler of investment in 2.0, especially in a weak global economy. Infoblox shifts the peak out further and gives CIOs the necessary breathing room to make the required new investments in innovation.
Cisco takes the stage with Infoblox because it makes business sense to address the antiquated core of the network as a means to trigger new investments in innovation. Yet Cisco and Infoblox are more than vision partners; that day they also announced a branch solution that enables more branch server consolidation and centralized management of distributed branch core network services.
When you mix in the strategic importance of application delivery, as networks continue to reach new kinds of endpoints and WAN delivery challenges, companies like F5 Networks (FFIV) and Riverbed (RVBD) and Blue Coat (BCSI) also evolve their solutions for the next peak. Connectivity Intelligence drives feedback loops between applications and infrastructure and breaks the hold of configuration and manual labor on IT investments in innovation.
The evolution of these leaders will fuel more opportunities for a new generation of cloud startups who address the next challenges. Former Cisco SVP Jayshree Ullal, the CEO at startup Arista Networks comments on a recent blog:
As 2009 begins, I have been thinking of how stale and stagnant the networking designs have been for the past decade, based on the classic three-layer of topology of core, distribution/aggregation and access. However, I do sense an architectural shift in the datacenter for cloud container environments changing the traditional constructs. Guiding this new cloud transformation is one of my major new-year resolutions.
At this point one has to wonder how the new IT infrastructure will take shape. Security pundit/blogger Chris Hoff touched off an interesting discussion last week about the future shape of the core of the network. Perhaps Chris intentionally or not explains Cisco’s interest in blade servers as well as core network service software on branch Integrated Service Routers (ISR):
A little light went on in my head this morning regarding how the cloud, or more specifically layers of clouds and the functions they provide (a-la SOA,) dramatically impact the changing landscape of what we consider "core infrastructure services," our choices on architecture, service provisioning, and how and from whence they are provided.
Specifically, the synapse fired on the connection between Infrastructure 2.0 as is usually talked about from the perspective of the evolution from the enterprise inside to out versus the deployment of services constructed from scratch to play in the cloud.
Disclosure: long Cisco