4 Mar 2013

Switching - Introduction

3 comments
Hello folks,
In the following articles we will talk more about switching, network implementations and wireless connections. In this first article, I will talk about the elements needed to consider in a switched LAN environment. The best way to implement a switched network is to design a hierarchical network model. There are three main levels of network functionality invented in order to provide a fully scalable network design:
- access layer - this layer is the one responsible with communicating with end devices. The switches used at this layer must provide the means for connecting desktops, printers, IP phones, etc. This is where the rules for providing secure access to the network ports are made.
- distribution layer - it acts as an intermediary layer between access and core layers. All data received from the access layer must then be passed to the core layer. Devices that form the distribution layer are usually more advanced switches that must provide high speed links, high availability and redundancy. This is where VLANs are configured for segmenting traffic into different subnetworks. We will talk in a future article about VLANs, about their role and functionality. 
- core layer - the backbone of computer networks. This is where traffic from distribution layers are aggregated and forwarded to external networks. The core layer is build from powerful devices that are capable to provide redundancy and high speed connectivity. 
   The network diameter is an element that measures the number of devices packets must travel inside a network. By considering this element you can avoid network delays.
Because switched networks evolved and new technologies appeared, today a switched network can provide different services over the same infrastructure (we call this a converged network). A converged network is a network that provides data, voice and video transfer between devices. By using a QoS service (quality of service), devices can transfer different kid of traffic over the same network connections (voice, data and video).
   This model is usually implemented in mid to large networks. The access, distribution and core network devices are usually stored in a special designed room called the telecommunications room. In the telecommunications room, switches, hubs and routers are stored in networking racks. These are special designed metal closets in which devices are stored one above the other just like normal shelves. The hierarchical network model offers some important benefits:
- bandwidth aggregation - device can use aggregated links between each other. This is a technique in which multiple switch ports act like one high speed link.
- redundancy - by using multiple routes between devices, if one link fails, traffic can be forwarded to backup links.
- scalability - networks are easily scalable.
- security - by providing the access layer in which rules for interconnecting end devices are made, these networks are more secured than others.
- maintainability - by using this three layer design, networks can grow without affecting maintainability.
- manageability - because hierarchical networks have the access, distribution and core layers separated, they are more manageable than normal network implementations. 
- performance - the performance is highly increased in these kind of network implementations.
   After implementing a network, special software tools and hardware devices are used to analyze how the network behaves. Traffic analyzers, monitoring tools are used to maintain a continuously working network. By using these tools, network Administrators can troubleshoot network failures. Another aspect that you must consider is how network devices and resources are accessed by other devices. Imagine that you have a company with different departments. A computer from the HR department must access all HR resources, it would be inappropriate to connect such a computer with a computer from the IT department. Consider how devices access servers resources in which data is stored.
   The next image will display a usual hierarchical network implementation:
Network implementation

   This is an example of a network topology diagram. These drawings are used by Administrators to better understand how devices communicate between each other. 
   Switches that are used in network implementations have different features and options. Some switches have fixed configurations and some have modular configurations. A modular switch offers a lot of flexibility, it can have multiple port cards, multiple power supplies and cooling systems. Fixed configurations switches can't be further customized. Multiple fixed switches can be stacked between each other by a special designed cable so that all switches are manged as one entity.You should also consider switch port density, forwarding rates and speed.
   I've written earlier that each layer have some special features as follows:
access layer:
- VLAN configuration - the ability to configure vlans
- link aggregation
- QoS (quality of service)
- Power over Ethernet - the ability to power up devices from a remote location
- port security - who can access the network.
- fast Ethernet/Gigabit Ethernet ports - specify the speed of a certain link
distribution layer:
- high speed
- link aggregation
- Quality of Service (QoS)
- Redundancy on both connections and components
- layer 3 supprt
- Gigabit Ethernet or bigger connections
- Access Control Lists - security configuration rules, we will talk about these elements in a future article.
core layer:
- Gigabit Ethernet or bigger connections
- High speed
- link aggregation
- Quality of Service (QoS)
- Redundancy on both connections and components
   I hope I've pointed out all the starting elements that you need to know when talking about network switches. We will continue discussing about switching technologies in the following articles. Meanwhile comment if you liked it or add any questions that you have, have a wonderful day.
author

About me

After finishing a Computer Science University I've started working as an IT support technician for an Outsourcing company. Since then, I've changed my profile a couple of times and now I work as a System Administrator at one of the biggest companies in the gaming industry. I'm constantly learning new things in this domain so I thought of creating this website for sharing my experiences. Until now, I've taken the Cisco Certificate Network Associate exam and several Microsoft certifications. Now, I am in the process of learning Unix and enhance my programming skills.

3 comments :

  1. Great article. However I would have to disagree with this statement "The best way to implement a switched network is to design a hierarchical network model". Although it maybe structured and easy to troubleshoot when a network is built using the 3 layer approach, latency related application will suffer. I work in an environment where every microsecond counts and having the servers directly connected to the core is the best way to reduce latency. We are concerned about 8us that is caused by each layer 3 hop therefore all our time critical servers connect directly to the core.

    ReplyDelete
  2. Thank you for your important post.Well explained, great tips and really all is awesome !!

    ReplyDelete
  3. Good advice. I'll take it into account. Thanks for sharing with us...

    ReplyDelete

Blogroll