Last week, Sun Microsystems (JAVA) Cloud CTO Lew Tucker predicted that IT expenses would increasingly track to the cost of electricity. “Lew’s Law” (as described to a room of thought leaders) is a brilliant theorem that weaves a microcosm of IT trends and recent reports into a single and powerful concept.
Lew’s Law is a powerful idea whose time has come, with profound and far reaching impacts, including the automation of the network. That automation will shape the fortunes of network equipment leaders, including Cisco (CSCO), Juniper (JNPR), F5 Networks (FFIV) and Riverbed (RVBD).
Those who see this trend coming and align their offerings with these new demands will obtain strategic advantage over the leagues of tired offerings continuing to monetize complexity and manual labor as CIO face increasing pressures from alternative cloud computing models and the spread of virtualization (more change, more sprawl [endpoints and servers]).
While Lew is focused on electricity as a cost, the power of his theorem is driven implicitly by the power of IT automation. For Lew’s Law to be prescient either electricity costs accelerate up and/or labor costs drop precipitously downward. For labor costs to drop, the ancien' regime of outdated processes and manual labor must be swept aside.
Here is his comment from the Infra20.com blog:
While I'm sure this is more obvious than brilliant, the cost of computing will continue to fall bounded only by the cost of the power to produce the computation. We're simply turning electricity into computation and communication using electrons to move around bits. What cloud computing does is to use automation, scaling, multi-tenancy, and a competitive marketplace to bring us closer to this lower bound for the cost of computing. Some may be concerned about this slide down towards commoditization, a collorary might be that this is good since cheaper computing can only drive up it's usage. Service providers such as Amazon and Rackspace are already seeing that while delivering the base-level computing plan is necessary their real sustainable value is in how computation is delivered, supported, and turned into higher value services.
While I'm sure this is more obvious than brilliant, the cost of computing will continue to fall bounded only by the cost of the power to produce the computation. We're simply turning electricity into computation and communication using electrons to move around bits.
What cloud computing does is to use automation, scaling, multi-tenancy, and a competitive marketplace to bring us closer to this lower bound for the cost of computing. Some may be concerned about this slide down towards commoditization, a collorary might be that this is good since cheaper computing can only drive up it's usage. Service providers such as Amazon and Rackspace are already seeing that while delivering the base-level computing plan is necessary their real sustainable value is in how computation is delivered, supported, and turned into higher value services.
It seems that one can forget the endless buzzword proclamations about cloud computing, what it is and who is delivering and instead focus on the ultimate rise of IT automation. Feel free to ask Lew directly, but I think he was suggesting if not predicting the ultimate automation of IT, the end of today’s tight link between manual configuration and needed outcome.
That automation won’t be trivial or easy but it is necessary for the evolution of IT into a more energy-efficient, powerful and strategic force in the global economy. It’s also necessary for the survival of IT, as network endpoints mushroom and new initiative pressures build, including virtualization. IT pros will need to evolve into experts on policy and strategy versus scripts and configurations.
If Lew was in a vacuum it might be easier to dismiss his prediction as far out, citing today’s legion’s of clerks, configurations, committees and procedural delays as ample evidence of an unchangeable status quo. Add a few fax machines to this manual labor empire and you have a business culture not unlike the legions of “middlemen” replaced by the evolution of the network over the past several decades.
Yet “the network is the computer”, “just-in-time IT” and “real time infrastructure” all align with Lew’s Law. Late last year Gartner published its first DDI MarketScope grouping appliances that automate DNS, DHCP and IP address management. Early in January Enterprise Strategy Group published a 2010 Outlook for the networking industry. I’ve pulled a few highlights that align with Lew, including the eventual alignment of networking, computing and virtualization:
According to ESG research, server virtualization is the number one IT spending priority for 2010. The networking industry will support this trend in several ways. Cisco’s whole UCS and Data Center 3.0 strategy is built on the assumption that networking, computing, and virtualization are coming together. Beyond strategy, Cisco’s Nexus 1000V and VN-Link products already align networking and virtualization intelligence. Expect other vendors to turn up the volume on similar initiatives in 2010. Juniper’s secretive “Project Stratus” will begin to bear fruit, as will comparable efforts from Extreme Networks, Force10, HP, and startups like Overlook Networks. ESG also anticipates more activity on the standards front to bridge networking and virtualization intelligence. (Jon Oltsik, Enterprise Strategy Group)
Some of the report’s suggestions might be seen as forward-looking, including the final chasm crossing of IPv6, but his perspective on F5 and the rise of network automation are very consistent with Lew’s Law and the recent Gartner MarketScope.
Check out Jon’s predictions regarding the overhaul of the network services at the core of the automation challenge:
Network services infrastructure upgrade. Imagine the future: massive data centers with tens of thousands of servers dynamically allocating processing loads to meet demand and conserve energy. Cool stuff, but how will this play out in the network? Even more fundamental, how will large organizations manage their IP addresses, provision them dynamically in real-time, manage ever-changing DNS, etc.? Many organizations manage IP addresses using spreadsheets and homegrown tools—there is no way this will scale when we need dozens of virtual IP addresses per physical device.
Jon understands the issues at the core of the network and the automation that must take place in order for networks to prepare for their ultimate support of IT automation. The rest of the ESG report contains more detail about how network automation will evolve and the increasing influence of virtualization on the network itself.
For virtualization vendors, it might be tempting to glance at this industry as an opportunity to build ever vaster partner ecosystems, which both VMware (VMW) and Citrix (CTXS) have done. Or they view the existing network as a new growth opportunity, with hardware increasingly standardized and software increasingly deployed on ever more powerful blade servers.
With virtualization the network terminates inside the hypervisor.
You can follow my rants in real-time at Archimedius.