Less is More
One of the challenging things about working with engineers, software developers, and really anyone whole likes to see their product be successful and grow ... is managing technology. To most people, whether its their home or car, success means BIGGER. But the challenge of technology is that it is getting more dense and performance is increasing at a rate much faster than the growth of most company's products and services. Another way of saying this is that the rate of change in the technology is greater than the rate of change (growth) of the product. For all but the very largest (hyperscale) .coms this is definitely the case. Likewise for most non .com enterprises, technology is improving in density and performance faster than business growth.
What used to require 18 servers today requires 1. What used to require 80 servers, now 4. What once required dozens of full cabinets of gear can now fit into a single flash shelf. A dozen fully loaded cabinets (48 RU) is easily over 100PB of flash storage today. Literally 250 times more capacity. What once required 200 Ethernet switches can now be done by 2.
What's hard for customers (be they internal or external) to accept is that their years of software development, application building, or infrastructure scaling, is now reduced to a single computer. Or more realistically to a fraction of a single computer.
Even if your company or operation does not use the cloud the concept does help people begin to let loose their bear hug on bare metals. Whether in the public cloud or in a private cloud, virtualization and container technologies let us not only un-couple the compute workloads from baremetal, but can offer us a way to provide decouple projects, applications, and even individual people from their beloved servers.
Reducing compute and storage space and power is absolutely critical to containing costs. Almost by definition, the vast majority of all compute and storage deployed is obsolete, power, and space inefficient. Even if you get the latest and greatest, something new is sure to follow with greater power and space density. While no infrastructure can be 100% up to date for more than a few months, on the other extreme running compute several years old or more is likely to cost more in power and space than the potential savings of refreshing the technology. The Return on Investment (ROI) is simple. If the capital cost of the equipment, as depreciated, exceeds the cost of the operation of the old gear (less the operational cost of the new gear), then you have a negative ROI and you should have (past tense) already refreshed. In other words, you are literally burning cash. There can be more complexity with tax implications, but if a flat line or simple (IRS) depreciation of capital shows the ROI in the lifetime, clearly there is a case for a refresh. But often if you just look at one system, the savings may not be evident. If you collapse 18 servers into one, it turns out you reduce your switch port counts, perhaps fiber or copper cable use, power outlets, etc. This is the trickle down savings. Looking upward, if you reduce your critical power consumption and space, you can perhaps shrink to a smaller space. Thinking bigger, replacing all of those spinning drives with flash means less swapping of drives and reduce staff time. Use that time to up-skill your staff, work on your disaster recovery, or come up with new ideas to try, test and build on.
If you shrink your footprint you will find that you will have more $ left for new projects, more space for new projects, more time for your staff to learn new technology, greater flexibility, higher reliability, and higher performance. The cost savings possible are incredible. There is literally a 95% cost savings with some infrastructure reductions if they are done simultaneously. For example, increase your compute density, convert to 40G/50G/100G Ethernet, reduce switches, and converted to converged flash storage -- all at the same time and the savings will literally be 85% to 95% of power and space.
Less is more. As much as the latest gadget is smaller, weighs less, and has more features, we should remember that the same is true in infrastructure. It time to stop building phone booths because the phones fit in our pockets. Stop building massive server farms when what is really needed can be a few VM's that fit in less than that magical 1.75" of one rack unit.