In the modern campus LAN, enterprise administrators are expected to provide a wide range of services with a lower operational budget than ever before. As a result, in order to balance competing requirements, administrators are forced to look internally to see how they can pour another pint into their existing half-pint pot.
Administrators are forced to look internally to see how they can pour another pint into their existing half-pint pot.
Network efficiency is one of the key mechanisms to achieve this, but where to start? Essentially, network efficiency falls into two distinct camps: absolute efficiency and resource efficiency. Absolute efficiency examines how an organization can meet its needs while it consumes the fewest resources possible (i.e., power, cooling and space). Resource efficiency encompasses the best way to deploy resources while reducing unused capacity (i.e., network interface, power supplies and CPU cycles).
Absolute efficiency is easier to consider at the design and build stage of any given campus infrastructure: Once equipment is placed into the rack, the facts are difficult to change. Resource efficiency is an ongoing process to ensure that what you have is used optimally. Sadly, one cannot wave a magic wand to improve efficiency overnight. But there are ways to improve your cost efficiency on the campus LAN. Here are five key steps to consider:
- Reassess your standards. What would do if you had a greenfield site? Assuming you are dealing with a continuous rollout, it is very difficult to break the blind inertia of deploying standardized equipment, laboring on the assumption that uniformity will at some point make your life easier. Too much rigidity in your standardization may result in your addressing requirements that never come or worse, deploying obsolete platforms. Network vendors constantly tweak their portfolios; staying on top of this is critical if one is to avoid the crime of deploying gargantuan routers in a tiny, remote location.
- Factor in issues that aren't your problem (yet). If you are responsible for any number of devices, all that gear is going to show up on an electric bill somewhere. Maybe not today, but someday soon a building manager is going to come to your door and "suggest" that you do something about your power draw or else find room for it in your (probably nonexistent) operational budget. Many organizations are coming under increasing physical pressures; an empty machine room that once housed an enormous mainframe 15 years ago makes little sense when now the rank and file are forced to sit two to a desk. Modern, physically dense switches require less space and cooling than the previous generation of tin. Consider what real estate you can trade for better equipment with a smaller footprint.
- Don't be afraid to rip and replace. Overcome your hoarding instinct. Striving for a more efficient network is not without its price, and you must critically reassess everything deployed -- from the core to the edge. If there is a better approach to your current requirements, you must pursue it. Avoid technologies marketed on the basis that they prevent future forklift upgrades; you are either paying for capacity you'll never need or adopting a technology that was legacy the moment you opened the box. Keeping a not-fit-for-purpose technology around on the off chance something changes is an easy way to tie up capital and operational expenditure indefinitely.
- Consider the management overhead and start to automate. It's tempting to stick with a single vendor or even network operating system if you are intimately familiar with its command line interface. However, making hundreds of small changes across an enterprise on a daily basis is a recipe for inefficiency and madness. Tools such as Puppet, Chefand Ansibleare now getting vendor attention as a means to automate network changes across platforms big and small. Time invested in network automation can radically reduce the effort expended on business as usual, and release you for more productive (and career-enhancing) endeavors.
- Assess, optimize, assess and optimize again. To get the best resource efficiency, it is critical to continuously assess and optimize your environment in a very tight feedback loop. Automation tools have an additional benefit in that changes are easily rippled throughout the network. Deploying a device once, then never touching it again is an admission that it is under-utilized and under-optimized. Resource utilization needs to be baselined and monitored continuously from the core, edge, WAN and to the cloud.
Millions can easily be spent on performance management tools for even a relatively small campus network: In my view this is anathema to the goals of efficiency. A more pragmatic approach is to make product selections based on specific pain points such as virtual server utilization, network capacity or incident monitoring. In most cases it is better to do one or two things very well rather than attempting to deploy a sprawling technology that tries to cover all the bases.
The march of commoditization has allowed advanced features that previously were available only from premium devices to travel to the entry level. This has made the job of selecting products based on absolute efficiency much easier, but often little thought is put into resource efficiency after the fact. Given the fact that in many organizations the selection of equipment is made by teams different from the ones that actually maintain it, it is perhaps not surprising that many are looking for efficiency gains from their network.
About the author:
Glen Kemp is an enterprise solutions architect for a U.K.-based managed services provider. He designs and deploys network and application security tools, including access control, remote access, firewalls and other "keep the bad guys out" technologies. He is also an experienced professional services consultant, delivering elephants and not hunting unicorns. His blogs can be found at sslboy and at the Packet Pushers Podcast. Follow him on Twitter @ssl_boy.