I recently had CAT6 network cabling run under my house and thus, the intent of this new article series is to share the experience and provide tips to others who are considering the effort.
An initial write up of the first article led me to realise that the amount of material here deserves multiple articles, and thus the birth of this new Writing Serial has eventuated:
This article focuses on the design considerations which I have separated out from the Initial Decisions article, which are more generic in nature. In this article, the focus is on the logical network design as well as the physical layout design activity that I undertook.
Logical Network Design
The traditional definition on the topic of logical network design revolves around the topology, or the way in which devices are logically arranged within the network. Historically, before the predominance of the Ethernet standard, there were token-based protocols that were complemented by a physical token/ring connectivity. Nowadays, the logical network design are focused more so on whether a network is client-server architected, or peer-to-peer (P2P). Generally, with Ethernet standards in force, networks are P2P in design; any device on a network can see another device and communicate with it.
For the purposes of a residential wired network solution, the majority of wired solutions tend to align with the star topology, where all end-point devices connect to a central infrastructure/access point. This same star topology is evident in the single point of failure that exists across both wired and wireless networks; the wireless access point/wireless router is the central point through which all devices gain access to the shared internet access. The same characteristic applies to a wired network solution, where the central hub/switch is the core point which all devices connect to.
Network End-point Design
The design of my home network is for six ports across three rooms/wall plate locations, with the lounge wall plate being a shared location for four ports. All six connections feed into a central wall plate, which is convenient since the wall plates can accommodate a maximum of six connection/keystone jacks. Thus the physical topology is a star model with the arrange of all devices connecting into the single hub/switch.
Initially, I thought that a hybrid design, where the four cables/ports to the lounge could be simplified such that I use a bullnose type wall plate which is designed so that many cables can pass through. After testing this design out with my Category 5 cables, which were used for the initial run, I decided on changing to a wall plate with keystone jacks finish instead. Since the positioning of the cables in the lounge is side by side to the surround speaker wiring/wall plate, this decision makes sense. At the central network hub/switch end, the wall plate with six keystone jacks also is much more presentable, rather than have six cables crammed through a bullnose wall plate.
Central Network Hub Design
The location of the central network hub/switch was briefly considered. As mentioned previously, the position of the main internet access point has shifted over the years. Originally, based on a dial-up connection technology, the key requirement for access to the internet was a phone line. In the first years, having an external modem meant moving that device along with the computer it was attached to. In the early 2000s, we migrated technology to hybrid fibre coaxial (HFC) cable modem – mainly because Optus was rolling out their network at the time and were offering bundle deals along with the home phone. Ever since, we have remained loyal Optus cable internet customers, even though we have threatened to migrate away on numerous occasions.
During the first few years of having a cable internet connection, we followed the previous principle and had the single connection split so that an additional two extension points were added to each of the bedrooms, with the exception of the master bedroom, which remained outside of the consideration set of potential computer desk locations. With the mainstream adoption of wireless network technology, the need to move the central internet access point was reduced to the point that the cable modem connection and central network hub has been stable at least for the last five years – fixed in the one particular spare bedroom.
Two years ago, I introduced the first serious network element beyond the basic home residential environment – a Netgear Gigabit Unmanaged Switch with 16 ports (GS116). Whilst the unit is not rack mountable, its simple design allows for easy transportation and relocation.
Relocating the central network hub, along with the cable modem away from its present location was briefly considered but not pursued for now. The idea however, remains valid as part of future scalability along with taking the networking effort to the next level. I will revisit this in the later article on future development opportunities…
The physical design and layout for the network cabling was a very important activity. The importance of working out the physical layout of where your network cables run, and the end-point locations of the various wall plates is essential to calculating the length of cable required. I actually had the opportunity to do a trial run with existing Category 5 cables.
In simple terms, I had earmarked two main locations for the cables to run under the house – one route would be shared by three or four cables, covering the home entertainment devices, whilst the other route would be common to two wall plate outlets that I had planned to be located at the same position of the one wall, just opposite each other in the two adjacent rooms. In this way, the different length of cable required would be simplified to two lengths.
Taking an existing layout map of the house, I measured all the walls to work out the maximum length of cable required. This is where things like mathematics has a practical application in real life! Instead of measuring the minimum distance/length of cable required by taking a direct route, I was measuring the maximum distance based on a longer route of following the walls. In this way, I estimated and cut three cable lengths to 10 metres, which proved to be way too long and I had over 2 metres of excess cable. Whilst I had used 10-metre cables for the home entertainment unit location, the other spare cables that I had were 5 metre lengths, and this proved to be slightly short of the required length for the two network ports in the bedroom locations. According to my measurements, 6-7 metres would have been a more accurate length.
Selecting the position of wall plates and the resultant cable length also required consideration of height in terms of how far off the floor would the wall plate and holes be positioned. The height factor was not as significant since I was merely following the existing standard utilised by the power-points around the house. An offset of 25 to 20cm in height for all the wallplates was measured.