System administrators have a full plate. Maintenance, monitoring, and management of their organization’s IT infrastructure keep them busy, leaving few opportunities to complete a thorough network design. A poorly designed or unorganized network, however, often requires more attention, and can be costlier down the road, making it worth the time investment up front.
If you have an opportunity to address the key requirements of your network infrastructure and organization as a whole, administration becomes easier in the long run. Here are three major steps for approaching network design to put you on the right path.
Gather initial network requirements
- Know your network. How many users connect to your network — 100 or 10,000? Do you have enough bandwidth to support those users? What kind of traffic profile are you looking at? Understanding the traffic on your network will help you make decisions down the road in terms of capacity and what protocols need to be addressed.
- Understand your organization’s expectations. What are the requirements for overall uptime for the network? Does your organization need three nines? Five nines? It doesn’t have to be exact, but you want to have an idea of what your business expects so you can design a network to support that. If your organization doesn’t require anything beyond two nines, putting in secondary power supplies would be foolish. At the same time, if your organization requires five nines, then not having the second power supply is just as foolish.
- Determine the budget available and how that fits your requirements. The right products for your network depend heavily on budget. The perfect network switch might be too expensive for some organizations, but they still need to choose the right switch family or product line. The switch might need to support certain features, like dual power supplies and layer 3 or light layer 3 protocols. It might need to do some kind of inter-VLAN routing, and should provide a command-line and web-based interface. If these requirements are missed initially, it’s almost impossible to get them later on after the purchase.
The perfect network switch is elusive. Organizations want a cost-effective option, as well as one that supports the majority of protocols, from RFCs to IEEE 802.1 standards. But the perfect switch must also be smart, simple to implement and maintain, and primed to adapt to new technology, all of which are more difficult to find. It must also be capable of providing an overall data center network topology that solves the organization’s need.
Here are five aspects of the perfect network switch that avoid the common challenges that network administrators typically deal with when implementing and maintaining network switches.
- Uses a command-line interface you’re used to. Most admins don’t want to relearn another command-line interface. The perfect switch offers an interface that’s similar enough to what you’re using and doesn’t require you to relearn anything. A gentle learning curve is an important requirement.
- Has a flexible product portfolio. Work with vendors that offer not one solution, but many different types of solutions. Organizations have different requirements based on what they’re trying to do with their network — cloud switches are different from data center switches and workgroup or office switches, and each type of switch has different needs. By working with a vendor that offers a variety of switches within a product group — from 1GB to 100GB — organizations can more easily support current and future requirements.
- Supports future technologies. You must identify new ways to connect switches to take advantage of all the connections they can support. For example, Spanning Tree Protocol (STP) has been widely adopted across most data centers by networking engineers, but there’s also a new one called TRILL that provides advantages with a greater number of links across all the switches, creating a fabric. And then there’s software-defined networking (SDN) — a topic suitable for a complete book. By supporting future technologies, the switch prepares for future requirements and enhancements to the overall network.
- Commits its configuration across the entire environment. When adding another switch to a network, the perfect switch should be able to download the current configuration for the network and apply only the parts that the switch needs for its own use within the topology. This includes VLANs, default settings, route tables, and other applicable networking settings. In a sense, the perfect switch eliminates the need to update every switch in the network. The network fabric does this work for the administrator instead.
- Integrates with and understands the hypervisor’s view of the network. There should be integration between the switch and the virtual hosts and virtual machines (VMs). When a VM moves among the various hosts, it can bring its VLAN or have its VLAN ready for it to use. The perfect switch is smart enough to do that for the administrator. This eliminates the time an administrator has to spend configuring all of the VLANs necessary on every switch or supporting changes.
It’s been just about a year since we rolled out the SHI Cloud, a milestone that has made us take a look back on the past year to see where we’ve been, what we’ve learned, and what we see for the coming year.
Since the SHI Cloud debuted, we’ve learned what our customers need from the cloud, how they use the cloud, and most importantly, how we can improve their experience in the cloud. Over the next few weeks, I’ll be sharing the five main lessons we’ve learned in the past year, as well as my predictions for the future of the SHI Cloud. Continue Reading…