In today’s fast-paced world, change is the only constant, and nowhere is this more evident than in the enterprise IT arena. The evolving nature of technology makes it mandatory for organizations to review and replace data center network cabling in order to cope with constantly growing business needs.
Data center network cabling is the backbone that supports your business needs. For appropriate data center network cabling, there are myriad designs and solutions to choose from. Use these tips to determine what’s best suited to your data center.
1. Appropriate cabling design
Data center network cabling design
Efficient cabling requires meticulous planning and design. There are five key functional areas in a data center, as stipulated by the TIA-942 standard:
- Entrance Room (ER) –The entrance facility to the data center.
- Main Distribution Area (MDA) – This area holds equipment such as routers, LAN and SAN switches.
- Horizontal Distribution Area (HDA) –The telecom room.
- Zone Distribution Area (ZDA) (Optional) –The consolidation point for all the data center network cabling and switches.
- Equipment Distribution Area (EDA) – The main server area where the racks and cabinets are located.
There are two main types of data center network cabling designs -- backbone cabling and horizontal cabling. Horizontal cabling supports mechanical termination, patch cord and horizontal cross-connect to the HDA, which may connect either to the ZDA or the EDA. Horizontal cabling can be fiber, copper or a mix of both. Backbone cabling on the other hand simplifies the design of the data center network cabling as it supports patch cord, mechanical termination and main cross-connects to various areas of the data center. This helps in better management. However, backbone cabling may face challenges in a distributed data center topology that supports multiple entrances.
Entrance Room > MDA > HDA > ZDA > EDA
A third possible design is zone cabling. With this approach you could have a consolidation area for all connections in the data center. You could either connect blocks or patch panels. The problem with this approach is that it may lead to a higher amount of moves, additions and changes, thus increasing costs. It is also difficult to design data center network cabling in the cross-connector channel with this approach.
Design your pathways keeping maintenance and changes in mind. You should be able to track and address the problems on the fly without impacting your production floor. For pathways, a top of rack (TOR) cabling pathway is recommended. However, improper pathways on top of the rack may impact heat outlets and distribution.
When you run everything in TOR, you will need to consolidate the cabling into one switch rack. Ensure that the end of rack (EOR) or the consolidation point is correctly designed. This is achieved by evaluating fiber and copper cabling necessities, along with choosing the right connectors and terminators.
2. Space planning
For space planning, most organizations consider either a central patching area or independent switch blades dedicated to each server cabinet. The former is preferable as it enables running of all the fiber or copper cables from the core HDA and EDA to the central patching area. Central patching eliminates complications in case of hardware changes as it reduces physical access to the systems. If your industry is governed by compliance and security related issues, this solution is perfect for your organization. You could also add an intelligent patching to the patching field to increase security by automatically monitoring and tracking moves. Moreover, all the purchased ports could be utilized. VLANs could also be used to segment your networks as necessary.
Opting for independent switch blades dedicated to each cabinet of servers may result in underutilization of resources. If the server network interface cards are not utilized, the adult ports become very costly and inefficient. For instance, if you dedicate a 48-port blade to a cabinet that holds only six servers with two connections each, then the cost of the other 36 ports and their maintenance is a waste. However, with a central patching field, the additional ports can be used as needed elsewhere in the network.
Another option is to have all server cabinets and switch cabinets in a row terminating to a single packing field further on rather than at a central location. You can then bring the core connections from the main distribution area to this packing field. If you are a service provider, this option may work well in your favor, as segmentation facilitates network administration.
3. Media selection
The devil, they say, is in the detail, and this applies to data center network cabling too. It is necessary that you review the type of cable required – copper, multimode fiber optic or single mode optical cable. Also evaluate the terminators that are required for each connection. As the hardware requirements of the data center increase, power and cooling requirements may rise. It is important that you consider and review these elements to avoid issues. Alien crosstalk (AXT) is one such problem that may occur due to negligence. (AXT is electromagnetic noise that can occur in a cable that runs alongside one or more other signal-carrying cables.)
Data center network cabling is based on simple yet intricate details that often go unnoticed. For instance, every data center professional knows that if a copper cable is bent sharply, it may impact the transmission. Can a good data center network cabling design prevent this? Yes. But is that ever addressed? Rarely!
About the author: Neeladri Bose is vice president sales (architecture services) at Cisco Services, India & SAARC. He has over 20 years of experience in the infrastructure IT consulting arena with specialization in data center efficiency and optimization, IT service management, PS service development and management.
(As told to Mitchelle R Jansen)
This was first published in November 2011