Efficient Data Center cabling needs to be planned and continuously improved in advance

Switching from a group of servers to another group of more powerful servers should be easy. In fact, this process is so conventional that enterprises cannot provide service interruptions for more than three hours. However, when the data center technician starts this process, they will find that the cables connected to the old server are not marked and not correctly installed on the device rack. When they try to disconnect the cable, they will find that some cables are connected to the broken plug, and in many cases, it will be hard to figure out where these cables need to be inserted into the new server.

Let's make a short note: the interruption would have taken 20 minutes, but it may take 5 hours. In addition, in the face of the fear of service recovery, technicians will concatenate temporary cables between equipment racks, which will be wound together and eventually, in order to clear these cables, the service needs to be interrupted again.

This is a real story. Sadly, the problematic servers are all key task servers. Their interruptions will seriously affect the enterprise's income-generating activities.

Data center cabling is a key issue

As we all know, wiring is an important part of the internal network of the data center. Enterprises not only need to know the whereabouts of each cable, but also need to be able to find specific cabling when devices are reconfigured or changed. This means that each data center should have a wiring list and wiring list management system. Data center administrators must be able to obtain detailed information about each cable in seconds, including the cable type and supplier, terminal contact, Cable Pipe location, and installation date.

The good news is that companies do not need to use workbooks to build such a system from the very beginning (although this is the most convenient way to start from the beginning, it is always better than not ). We have various cable management systems, from the most basic to the most complex. Some systems are also bound to the network management system to provide real-time access to the inventory system. When a network device fails, this can help solve the problem instantly. Many other systems are part of the big data center infrastructure management (DCIM) application. For example, IBM and Rackwise provide a wide range of products in this field. There are also open-source solutions, such as openDCIM, which can also be used for wiring from the beginning.

Cable management strategy

First, the enterprise must mark the cables. At least, each cable terminal needs to mark the unique identification number of the cable and its purpose. Remember, when a fault occurs, it is important to be able to quickly and clearly identify the specific cable.

The data center should also have a specific cabling architecture: This architecture will allocate specific cable pipelines to specific features. For example, the interconnect cables between routers, switches, and firewall devices should be separated from the cables used for external communication. (I think of a data center like this. One of the twisted pair wires hangs on the ceiling and is used for external customers to access the auto-service quote processor. Every time the fan is turned on, this twisted pair will swing in the wind, and the customer's access will be interrupted ). In addition, the fault-tolerant architecture is also important, especially for crosstalk, noise, and heat dissipation.

Does it require a lot of work? Is it necessary? You can calculate that the custom cable may cost several hundred US dollars depending on the length. However, cable failure may result in loss of thousands of US dollars, or at least negatively impact productivity, because it is difficult for technicians to find a cable to replace it. A good cabling plan is like an insurance policy. When a fault occurs, it can instantly provide assurance for the enterprise.

So, how do enterprises know whether the data center is in danger? Visit the data center and check the wiring. Is the cable clearly labeled and color coded? Are the cables neatly arranged and isolated according to functions? Is the cable on the ground? If the answer to all these questions is no, there will be problems in the data center.

Check, Mark, and name

Where can I start? It is best to start by checking and marking each cable. If you need the wiring naming method, there are many practical guides on the network to help you develop labeling rules. After all cables are located and labeled, the company needs to enter cabling information to the inventory system or workbooks, although this is not the best practice. In the end, enterprises will want to have a more powerful system that allows them to increase, move, and change wiring without increasing the overhead.

Finally, after the inventory is complete and the cabling has been confirmed, the enterprise needs to continuously test and re-route the cables to match the functions and types. The improvement process should be continuous, which can significantly improve the cabling infrastructure in a very short period of time.

The option here is simple: either try to clean up data center cabling, or pay for inefficient or poorly labeled cabling in the future. In general, if you wait for processing, the cost will be higher, and this delay will have a bad impact on your business.