Software-defined architecture enables network optimization for cloud access

Everyone is talking about the huge changes that cloud computing will bring to the industry and how cloud computing technology will develop to provide unlimited applications, data and services to end users. However, as long as there are bottlenecks in the network, the efficiency of cloud computing architecture cannot be maximized. Many companies have rushed to transform to a cloud computing model without fully considering the limitations of traditional networks on overall application performance, which is quite unwise.

Simply purchasing bandwidth to solve network problems as in the past is no longer suitable for today's business status, especially when it comes to hybrid and public cloud deployments. Instead, a new network optimization approach that focuses on the application itself is gaining favor with enterprises. This article will introduce how to modify the network to help optimize cloud application and data access.

[[209031]]

The most popular approach is to relieve network congestion for public and hybrid cloud resources to strengthen the Internet and WAN edge. WAN optimization technologies such as compression, local content caching, and manipulation of TCP window size have always been the preferred technology for accelerating remote cloud access.

But with the emergence of SD-WAN technology, the industry seems to have found another way. Software-defined wide area network (SD-WAN) provides intelligent routing capabilities using two or even more paths to reach remote destinations. Path selection is based on continuously calculated variables such as available link bandwidth, minimum latency and minimum packet loss. This technology helps connect a company's branch offices to the cloud.

In situations where remote users connect to an office or cloud service, network architects need to consider different approaches. Before the rise of cloud computing, applications and data resided in private data centers or branch offices, and remote users typically used VPN technology to securely connect to a company's main office.

However, now that applications and data are distributed among various cloud service providers, it no longer makes sense to block remote access VPN tunnels for users at corporate headquarters just to send information back to the Internet or over expensive WAN connections. Instead, network architects should strive to simplify data flows and reduce the number of network hops as much as possible, which usually means establishing simplified secure access directly to public cloud resources over the Internet using IPsec VPN, SSL VPN, or secure Web access using HTTPS.

In addition, if the enterprise's users are spread all over the world, then developing a globally distributed cloud architecture is more appropriate for latency-sensitive applications. Globally distributed cloud deployment speeds up client-to-server deployment, and users can access resources that are geographically closer to them.

Today, network engineers need to have a deep understanding of how critical applications perform end-to-end. It is important to understand how data flows between servers and end users, as well as between servers and other distributed resources or resources that are not in the cloud at all.

With a clear understanding of data flows using modern network performance management (NPM) tools, network engineers can optimize resource placement to avoid network bottlenecks or congestion. It also helps enterprises choose the quality of service (QoS) marking and policy enforcement to determine the right priority for business-critical applications.

While all of the above network optimization techniques help improve cloud access, the ultimate goal of network professionals should be a complete end-to-end software-driven network intelligence architecture. Unfortunately, from an implementation and cost perspective, it is very difficult. This is why many companies will focus on solving specific areas of the network, such as cloud access optimization problems in WAN and data centers, but the industry's ultimate goal should be to use software-defined architectures to deploy optimization strategies throughout the network.