Seven factors to consider in network redundancy design

[[433681]]

【51CTO.com Quick Translation】 When a network fails, many businesses come to a halt. For some, a few minutes of downtime doesn't matter. For those businesses that are customer-facing, a few minutes of downtime means a lot of lost revenue and possibly even lost customers. For these businesses, it is critical to design a network that can remain operational even when components fail.

For enterprises where even a short outage would have a significant impact, adding redundant equipment and signing redundant service contracts is a must. However, adding network redundant equipment will increase cost and complexity, so the cost of downtime must be weighed against the cost of adding equipment and services.

Here are seven factors your network team needs to consider when designing network redundancy:

1. Switches and routers

Switches and routers are generally very reliable, but they sometimes fail. Some businesses keep an extra switch or router in a rack so they can quickly replace a failed device. Businesses with more critical needs must have redundant equipment up and running in their networks.

2. Network Protocol

Network standards bodies develop network protocols that can quickly switch to backup equipment in the event of a failure. Adding redundancy at the second layer requires IT teams to connect multiple switches to each subnet segment.

But these redundant switches can cause network flooding by creating multiple paths through the network, with multiple copies of each packet. Spanning tree algorithms provide a way to determine a single path through the network. Unfortunately, spanning tree can take nearly a minute to determine a new path. While this time may be acceptable for some networks, other networks require faster recovery.

Today, new protocols such as Multisystem Link Aggregation (MLAG), Transparent Interconnection of Lots of Links (TRILL), and Shortest Path Bridging (SPB) have been developed to support faster recovery. Network teams that design network redundancy and require faster recovery must determine which option is better for their network.

Seven factors not to forget when adding redundancy to your network design

3. Subnet connection

The next step in adding redundancy is subnet connectivity. Again, it is necessary for an enterprise to provide multiple paths between subnets. Routers connect subnets within a network to external destinations. Each subnet must be connected to multiple routers to provide redundancy. Protocols such as Open Shortest Path First (OSPF) and Enhanced Interior Gateway Routing Protocol (EIGRP) define how routers inform each other of the current best path to each destination.

A router can determine that a neighboring router is down when it has not received updates from that router for a few seconds. However, both Open Shortest Path First (OSPF) and Enhanced Interior Gateway Routing Protocol (EIGRP) require more time to recover than is acceptable for some networks. Hot Standby Router Protocol (HSRP) and Virtual Router Redundancy Protocol (VRRP) can be used to reduce the time required to recover from a router failure.

Network connections can be disconnected for many reasons - for example, someone unplugged the wrong cable, or a new cable became loose when adding it, or accidentally bumped against a cable when moving the device. IEEE802.3ad link aggregation defines how two network cables can be used as a single connection. As long as both are connected, traffic can be shared between the two cables, but when one cable is disconnected, its traffic is not interrupted.

4. Backup

A disk failure that results in data loss can paralyze a business. For some businesses, backing up once a day is sufficient, while others cannot afford to lose a day's worth of data and the time required to recover.

Redundant Array of Independent Disks (RAID) protects against the failure of one disk and offers several levels of protection. Each item of data can be written to two different disks. Higher levels of protection define methods of adding additional disks and parity information, which enables IT teams to recover the correct data in the event that two disks that should contain the same data are different.

Continuous cloud backup is better than the highest RAID level because sending every data update to the cloud means no data will be lost. A single RAID disk is installed in a cabinet, which means that if the cabinet is damaged, the disk inside may be damaged. At the same time, users still need to back up the array regularly, but the data accumulated since the most recent backup may be lost.

5. Processor

Processors can fail like other components, so the possibility of failure must also be considered in network redundancy design, and processors must also be regularly updated with the latest system software versions. Enterprises need to have sufficient additional processor resources to ensure continuous operation of the network.

Moving all processing and storage to the public cloud simplifies the task of redundancy design. There are many processors and storage units in the cloud platform, and if an application fails, it can be quickly moved to another processor for processing, and redundant storage can be configured. If an event occurs that causes the entire facility to shut down, the processing process may be transferred to other cloud computing infrastructure.

6. Power supply

Obviously, IT equipment cannot work without power, and power to a data center may be interrupted due to storms, a utility pole being knocked down by a car, or other unexpected events. In the event of a failure, the backup power supply of the data center can quickly take over. For large data center facilities, a proper UPS power supply is required.

Since switching to diesel generators takes some time, if the duration of the power outage exceeds the battery capacity of the UPS power supply, the backup diesel generator can be started to provide power to the load. In some cases, the data center can also be connected to two different power lines to prevent power outages due to damage to one of the power lines.

7. Wide Area Network and SD-WAN

WAN connectivity has always been important, but the growth of cloud computing and the importance of remote users has made WAN reliability increasingly important. One option for enterprises is to obtain connections to two different network service providers. While this increases operational expenses, it protects against failures in the link connecting to the network service provider as well as failures within the provider's network.

Software-defined WAN (SD-WAN) provides an additional way to add network redundancy. Multiprotocol label switching (MPLS) lines are very reliable and can guarantee specified quality of service (QoS) levels, but they can fail. SD-WAN controllers can switch traffic to the public internet in the event of a failure. The public internet does not provide the same level of reliability or quality of service (QoS) guarantees, but it provides a way to get data to its destination. Another advantage of SD-WAN is that it can shift less important traffic to the public internet during load peaks, rather than prompting IT teams to order the highest MPLS bandwidth levels.

Adding redundant equipment will increase expense and complexity, and network designers should not design more network redundancy than necessary, but they should also not design less than required because even a brief interruption can determine the success or failure of a business.

Original title: 7 factors to consider in network redundancy design, author: David Jacobs

[Translated by 51CTO. Please indicate the original translator and source as 51CTO.com when reprinting on partner sites]