Keith Bromley
Sr. Manager, Product Marketing at Ixia

How To Reduce Bandwidth Overload At The Edge

September 10, 2019 by Keith Bromley

There is a fundamental shift currently happening in operational technology today—the shift from core computing to edge computing. This shift is being driven by a completely massive growth in data. According to Cisco Systems, network traffic will reach 4.8 zettabytes (i.e. 4.8 billion terabytes) by 2022. Businesses cannot continue as usual and still keep up with network performance, security threats, and business decisions. So, in response, network architects are moving as much of the core compute resources as they can to the edge of the network. This helps IT reduce costs, improve performance and maintain a secure network.

A substantial contributor to this edge computing shift is the expense of multiprotocol label switching (MPLS) circuits, especially for the last mile to interconnect with remote offices. Network managers need to reduce their infrastructure costs and one to do this is to filter and groom the data as close to the edge as possible. To accomplish this, MPLS circuits can be shifted to Internet-based network transport links. This can reduce remote office interconnect costs while also increasing data bandwidth with 100 megabit per second (Mbps) and Gigabit Ethernet links that are readily available through ISPs.

Using this strategy, MPLS circuits that cost $200-400 per Mbps can be replaced with IP-based links that cost between $2 to $15 per Mbps. SD-WAN services can then be layered on top of these IP-based links to guarantee performance, as Internet-based services can introduce latency and packet loss to the network. Cloud services can also be deployed to the edge of the network.

Throughout this conversion, security and performance will be key factors. Security is an especially important concern as the network extends further out. Inline security appliances and SSL decryption are necessary to protect against the introduction of malware into the corporate network.

For this transition to be successful, a layer of abstraction between data access and analytic tools to drive rapid business decisions will be required. This abstraction layer provides the flexibility to meet the ever-evolving needs of the data center. According to Bob Laliberte, senior analyst at Enterprise Strategy Group, “Today’s network and security operations teams are faced with increasing data volumes and velocity that are exceeding human scale. It is impossible to effectively manage the security and monitoring of these highly distributed environments without an abstraction layer provided by a sophisticated network packet broker.”

In regard to network monitoring, here are some basic benefits from using IP-based links as far as possible to the edge of the network:

  • Protect core and distribution centers from data overload
  • Reduce network costs with a pay-as-you-grow model
  • Perform data filtering as close to the network edge as possible

A visibility architecture is the perfect way to create the needed separation and makes it possible to manage the data.

A visibility architecture contains three basic components:

1. Access layer

2. Control layer

3. Monitoring tool layer

The access layer of the model is concerned with access to monitoring data. A network device, whether it is a physical tap, virtual tap, or mirrored port, provides the access.

The control layer uses a network packet broker (NPB), which allows you to filter, aggregate, and load balance data. Packet brokers provide several high-level benefits, including connectivity, cost-effective security and monitoring tools, scaling, reliability, and longevity.

Visibility Architecture

The monitoring tool layer consists of the monitoring tools that receive filtered data from the NPB that is relevant and concise; rather than receiving the data directly from a tap. NPBs augment your monitoring tools to make them more efficient; they are not a replacement.

Ixia’s vision of an abstraction layer centers around the Vision X NPB and its modular design. Customers can select different functions, capabilities, and speeds as their data centers evolve. Network operations teams can upgrade the speed and bandwidth of their solution as needed without having to replace existing monitoring and security tools.

Here is a picture of a visibility architecture using taps for remote access which are then connected into successive packet brokers for aggregation.

Edge Computing

If you want more information on this topic, try reading this solution brief Reduce Bandwidth Overload with a Data Abstraction Layer.