To build an affordably scalable data center network, we believe that disaggregation with Cumulus Linux is the ideal networking solution (for obvious reasons). If you haven’t made the jump to open networking, we recommend you read our total cost of ownership (TCO) report, which covers how our customers saved up to 60% by making the switch. For the rest of you that have already come to the light (we prefer calling it the green), we’ve put together a few tips for optimizing your data center network for increased efficiency and lower data center TCO.

  1. Optimize your cables and optics
    The cost of optics varies greatly. If you’re running high-cost optics, we recommend doing your research and seeing if there is a lower-cost option that will suit your needs. As switches become denser, the optics and cables can actually cost more than the switches themselves — especially with proprietary, locked-in options from traditional vendors. Most traditional vendors source their optics from third-party manufacturers and then add a premium. When open optics are sourced directly from ODMs, you can reduce the cost of cables and optics by 50% to 70%.
  2. Automate like a boss
    Most organizations are doing some form of automation. With Cumulus Linux, you can leverage automation tools that you’re already using (like Ansible or Puppet) on your servers, and automate throughout the entire stack. The more you automate, the more you increase efficiency and the more you lower data center TCO. Plus, no matter what you’ve already automated, you could probably be automating more. For instance, are you automating all of your testing? This post tells you why you should. There are tons of ways you can optimize your network with automation. With Cumulus Linux, it’s easier than ever. Check out our technical documentation that covers all things automation. We bet you’ll learn a lot.
  3. Prepare for mistakes
    Expecting both your technology and your resources to run without error is a recipe for disaster. Mistakes will happen. That’s the reality of a data center (but you already know that, right?). The key is to put systems and processes in place to ensure you can fix errors quickly. With Cumulus Linux, you can leverage snapshots and rollbacks. If a command error is made, you can quickly and easily roll back to a previously known good state. Plus, you can leverage snapshots and rollbacks to build a strong redundancy plan, combining this feature with remote logging and a comprehensive monitoring configuration.
  4. Break silos
    Communication is key in all relationships, even in data centers. If you have completely siloed teams, or worse, individuals, then you have inefficiencies in your data center. That’s just a fact. Cumulus Linux is designed to optimize operator-to-switch ratio, so you can maximize talent across different demands and processes. If you see weaknesses here, this may be a wonderful time to evaluate and ensure your team has the resources, training and processes they need to efficiently succeed. This is where NetDevOps comes in. In short, you want to ensure you have clear, effective processes in place for collaboration, skill sharing and task sharing between job titles, and standardized tools across the stack. With Cumulus Linux, you can standardize easily and run the same applications you run on your servers. Need ideas on how to do this? Contact your sales rep to learn how to build better processes and systems.
  5. Be energy efficient
    When you think energy costs, the first thing that comes to mind is likely rack cooling. Great — making sure your racks don’t overheat is a wise way to prevent extra energy spending and thus lower data center TCO. But there are a lot of other ways to ensure your data center is energy efficient. Although a bit old, this post over at InfoWorld cites 8 great ways you can reduce energy costs in your data center by simply being aware of how you’re wasting energy. They might not all make sense for your business or building, but at the very least, take some time to evaluate your energy bill and wasteful hardware. You could likely make a few easy changes (like shutting down servers that aren’t in use) that will add up over time.

With the right hardware, software and processes in place, you can run a fully-streamlined, efficient data center network that is built for scalability. Of course if you need help coming up with solutions for improved efficiencies and lower data center TCO, our knowledgeable services team is here to help. Simply contact us, and we’ll connect you with an industry expert to help you design your optimization plan.