We began a two-part series last week with an examination of cabling and connectors. This series developed out of my own confusion between what I thought was an RJ45 and what turned out to be an 8P8C connector. In today's high-speed networks, the selection of the proper components is not only important--it's crucial to reliable operation. Let's continue the series, written by Alan M Frank, senior systems engineer, out-of-band access and technology strategist at OOBAXS
Cabling standards for CAT6 CAT6a really need to be followed in order to maximize any investment in a new high-definition, high-speed HD-KVM solution.
Standards for cabling are still high, but up till now a slightly looser and perhaps more casual approach has been taken overall. Let’s face it, after installation and certification, in the perfect environment, there is just no match for the day-to-day operations that can truly find the weak points (like being under load). Putting a synthetic load on a network for a few days before you roll it out, but who takes the time to do this these days in a data center (let alone for a KVM solution)? Managers often consider such delays and tests to be a luxury, and may not budget for them. Warning, this could be a critical mistake.
We’re only creating this network for a KVM Solution! Have you ever heard this statement? While one used to be able to install a lower-performing network for KVM tasks, that's no longer the case.
Today’s high-definition KVM solutions require sustained high-speed performance or they will not meet the needs of a broadcast operation. It’s one thing when you implement a network for the corporation to carry production traffic or for an out-of-band management infrastructure; as they both transfer packets and some days appear to be working properly. However, you may often hear someone say that the network is running slow today, which could actually be an abnormal amount of traffic, collisions, or CRC errors; it’s probably not a crisis because the files transfer completely, or the process the user is running eventually ends in nanoseconds; it just seems like a minor glitch, and is not truly a crisis. So, the underlying issue may ultimately go undetected, and this is where critical problems could be festering, waiting to rear their ugly heads at the most inconvenient, and costly times.
This would not happen in a network that supports an HD-KVM solution or in a corporate scenario where you depend on a robust environment to carry your day-to-day traffic.
If people are browsing an internal resource, and it slows down, it is just is a little inconvenient. If they are copying or transferring data, perhaps it may take a second or two longer. So, while the operation of the corporate business network is crucial things don't shut down even if packets are completely dropped. CRC-Errors are common, frequent, and often due to inferior cables, a loss of continuity, or COLLISIONS, and are CSMA/CD networks (Carrier Sense Multiple Access / Collision Detection).
Typical HD-KVM solutions now run at sustained packet rates in excess of 700Mb/s, which is highly demanding and stresses a network to its limits, constantly and consistently. When the user experiences a delay or dropped packet in this scenario it displays itself in ways that anger the users and workflow can slow. This is why a separate and isolated network should always be used for HD-KVM solutions.
As we know, video traffic is the largest consumer of network bandwidth; the higher the resolution and frame rate, the more bandwidth is needed. As a result, HD-KVM solutions demand high throughput over an isolated networks. This is why I recommend always splitting the HD Matrix traffic off to its own network.
If you are using a KVM solution that relies on an over-the-counter GigE switch on these networks, pay attention to handling this traffic properly for design purposes of the backbone, as well.
Because the requirements you build on are sustained, investing in a higher volume, quality assured product with sustained throughput across the backbone is a must. It’s important to pay close attention to the design of the backbone, using 100% utilization, and plan for adequate growth.
Alan M Frank is the senior systems engineer, out-of-band access and technology strategist at OOBAXS (pronounced Out-Of-Band Access)
For a thorough checklist of the visible signs to watch for in your KVM networks, visit: http://oobaxs.com/101checklist.