No Jitter is part of the Informa Tech Division of Informa PLC

This site is operated by a business or businesses owned by Informa PLC and all copyright resides with them. Informa PLC's registered office is 5 Howick Place, London SW1P 1WG. Registered in England and Wales. Number 8860726.

3 Ways to Improve Availability of First, Last Mile

AdobeStock_232447294.jpeg

An image of people connected through the Internet
Image: Sergey Nivens - stock.adobe.com
Broadband Internet has become popular for first- and last-mile enterprise WAN connectivity, i.e., the connection from the branch or HQ into the backbone and the connection to the data center or public cloud provider. Significant factors are motivating this transition, including the cost of MPLS circuits, improving the quality of broadband Internet links, and the emergence of cloud architectures and applications. However, using broadband Internet means you must manage the first and last mile and requires mitigation to address congestions or packet loss to ensure peak application performance.
 
According to NCTA, the Internet & Television Association, Internet speeds have grown from 50 Mbps in 2009 to over 2 Gbps in 2019, and we are looking at a 10 Gbps future. Considerable improvements in the available bandwidth have prompted enterprises to deploy this technology for first and last-mile connections. And, many bandwidth-hungry applications like large file backups and transfers benefit from the available bandwidth. Still, apps like voice over IP (VoIP) and video that are sensitive to delays and latency have a tough time coping with unoptimized internet transport, and due to this, it’s common for unified communications as a service (UCaaS) applications to experience dropped calls and poor call quality. Deploying a last-mile MPLS circuit to carry UCaaS traffic isn’t a great solution due to the expense, management overhead, and in some cases, the lack of flexibility. In many cases, the location of a branch office means MPLS isn’t even available as an option.
 
A graph on Internet speeds

Source: NCTA – The Internet & Television Association

The solution is to better understand the limitations of the broadband Internet path and implement link assurance technologies that will solve first and last-mile availability. Path replication, adaptive loss recovery, and load balancing are three potential features that help achieve this goal. These are in addition to the various higher-layer optimization technologies such as compression, transport optimization, data de-duplication, and even application proxies that also operate across the first and last-mile or end-to-end.
 
Path Replication
Enabling path replication can be considered as a loss recovery mechanism. Path replication allows customers’ selected traffic to be replicated over the primary and secondary links. If one of the links experience loss, the second link delivers the same packet to the peer by delivering duplicate packets between the edge and the node. Path replication can be applied to any kind of quality of service (QoS) traffic.
 
A graph showing Internet connectivity
In the image above, the HQ is connected to the SD-WAN service node over Internet links from two different ISPs. Packets are replicated on both ISP links. At the node, the packets are reassembled by combining both links, thus ensuring that any performance degradation or packet loss in one ISP doesn’t affect the connectivity.
 
Advantages of path replication include:
  • Substantial gains in application quality for VoIP for example
  • Elimination of variations in network characteristics, such as packet loss and jitter on a specific link
  • Elimination of packet loss and improved reliability
 
Adaptive Loss Recovery
Adaptive loss recovery is a feedback mechanism that selectively retransmits lost packets on a link. This may be implemented using a lightweight TCP-like algorithm to recover from packet loss. When the receiver, in this case, the SD-WAN service node or SD-WAN edge doesn’t receive a packet; it initiates a Negative ACK, requesting the peer to retransmit the lost packet. Hence, the amount of bandwidth required only needs to be proportional to the loss percentages over the links.
 
A graph on Internet connections
Advantages of adaptive loss recovery include:
  • Elimination of packet loss for better application performance on a single link
  • Increased reliability over a single link
  • Low bandwidth overhead and more efficient bandwidth usage
 
Load Balancing
Load balancing allows customers to take advantage of both paths, primary and secondary, by sending alternating packets down alternate paths. The traffic is distributed across the links on a per-packet basis in a round-robin manner, resulting in very efficient resource utilization.
 
A graph on Internet connections
Advantages of load balancing include:
  • Performance improvement as a result of distribution of traffic across links and avoiding traffic overload on a link
  • Better link utilization independent of the number of packet flows
 
UCaaS – Deriving the Most Benefit
Though these features improve the performance of an entire gamut of cloud applications, the ones that benefit the most are UCaaS applications. Voice and video are very susceptible to latency and delays. Besides, any disruption to the link results in dropped calls; re-establishing the call takes a few minutes if not seconds. This delay affects the user experience and perception of quality.
 
By deploying a cloud-first WAN that includes provisions for path replication, adaptive loss recovery and load balancing, enterprises can substantially reduce costs, reduce the limitations of the Internet path, and improve connectivity between the cloud and cloud-based applications — all while greatly improving application reliability and the end-user experience.