By: Teresa Hoffman, Fiber Optic Product Business Manager
Let’s face it. Data center upgrades are expensive and risky. There are countless factors to consider, and more than a few mildly terrifying things that go sideways if you make the wrong move. In some ways, you have to be a fortune-teller. You need to be sure that the infrastructure you install will be able to support not only current needs, but also your next generation of equipment and ever-increasing bandwidth requirements. The last thing you want to do is install a cabling plant today that you’ll have to rip out and reinstall three years from now.
I’m not going to attempt to cover all of factors that play into a data center upgrade. There are entire books, websites and even college courses devoted to that. But if we boil it all down, we’re really talking about two main considerations – budget and performance.
In the data center world, there’s more than one type of budget to consider.
Of course, there’s always a financial budget to keep in mind. We all want to have “the best” system/product/solution, but financial realities demand that we carefully examine what we need, prioritize the requirements, and choose solutions that provide the best return on our investment.
For example, looking just at cabling, traditional cable plants have used OM3 multimode fiber. Going to OM4 – or OM4+ – is more of an investment, but worth it because the added bandwidth might be needed to overcome insertion loss. Another example is single-mode vs. multimode. Single-mode is the least expensive cable, but with higher connectivity and optics costs, single-mode as a solution can be up to three times more expensive than multi-mode.
Financials are important. But they’re not the only budget you need to think about your migration path. What good is saving money on a system that doesn’t work? It’s critical to understand whether or not your cabling will support your protocol. To do that, you need to know your link budget – or power budget – and understand what you can do to maximize it.
In theory this is simple. Your system either works or it doesn’t, right? Not exactly. With the number of PMD options constantly increasing, it’s get difficult to keep track of all of the variables. A PMD is the Physical Media Dependent form factor for the transceiver connection. Each PMD has a set of conditions – fiber type, connector type, bandwidth support and cabling insertion loss – that are needed in order for it to operate correctly.
IEEE defines worst case conditions to ensure interoperability, but these limits can be very restrictive. As an alternative, many network managers take advantage of engineered links to improve performance. An engineered link is one where one or more performance attributes are specified to be better than the standard. The better performance can come from any number of different attributes. It could be from higher bandwidth fiber, or lower loss connectivity, or better performing transmitters and receivers. With the number of potential options available, the number of possible solutions seems infinite.
So now that you’ve factored in all of your financial and link loss requirements, how do you go about determining the best migration plan – one that balances your budget with your desired network performance objectives? Part 2 of this series will discuss several approaches to this challenge.