How Can Organizations Improve Network Performance?

June 1, 2020

Often, when speaking with network managers responsible for infrastructure within a multinational or global enterprise, I hear first-hand accounts of the impact of sluggish network performance. For example, videoconferences between engineers and product managers on separate contents can be brought to a standstill because of packet loss or latency. Similarly, slow networks can lead to painfully slow file transfers for large media files or CAD (computer-aided design). Further, poor network speeds can limit an enterprise’s ability to use cloud platforms to their full potential.

These conversations invariably wind up in the same place: how can the modern digital business improve network speed? And what does that look like in practice? Here, we’ll explore just that.

Top Five Ways to Improve Network Performance

Reduce latency, add capacity, and/or compensate for jitter and loss are obvious high-level answers to most WAN optimization challenges, but doing so effectively is where the real challenge lies. For the modern WAN, just throwing money at the problem and buying more capacity or more expensive network gear isn’t always the right answer. That means understanding the underlying problem (beyond “the network is slow”) and solving for that.

#1. Improve Middle Mile Performance

When MPLS (multiprotocol label switching) was the de facto WAN connectivity standard, enterprises often had a reliable, albeit expensive and inflexible, middle mile connection they could count on for enterprise-grade connectivity. However, as cloud and mobile grew in popularity, the inflexibility and cost of MPLS began to drive enterprises away. For example, the trombone effect (the inefficient backhauling of cloud bound traffic through a specific network endpoint) often meant MPLS connectivity to cloud assets was worse than standard Internet connections.

As a result, businesses turned to SD-WAN and Internet-based VPN solutions as an alternative. Unfortunately, because of the well-known problems with the public Internet, this meant an increase in latency across the middle mile.

The solution? An approach that provides the reliability of MPLS across a private backbone while also offering optimized connectivity for cloud and mobile. This is exactly what Cato Cloud was purpose-built to do. With a global private backbone supported by a “five nines” (99.999%) uptime SLA and strategically placed PoPs (Points of Presence) around the world (many sharing a datacenter footprint with major cloud service providers), Cato can provide reliable, low-latency middle mile connectivity without sacrificing the flexibility of SD-WAN.

#2. Optimize Cloud Connectivity

The cloud is ubiquitous within modern digital businesses. With more and more critical workloads being shifted to the cloud every day, the importance of fast and reliable network connectivity is growing. We’ve already alluded to the challenges MPLS and the public Internet pose to the WAN in general, and they become further magnified when you take public cloud services into account. In many cases, enterprises are turning to expensive premium solutions like Azure ExpressRoute or AWS Direct Connect to optimize cloud connectivity. The idea is simple: a direct connection to the cloud data center overcomes many of the network challenges related to accessing cloud assets. However, many platform-specific solutions cannot account for all the cloud workloads within an enterprise. Email, CRM (customer relationship management) software, and collaboration tools may all come from different cloud service providers.

This is why a solution that bakes cloud optimization into the underlying network infrastructure is important. For example, with Cato Cloud, enterprises can eliminate the need for costly premium solutions and provide an agentless integration to connect to cloud datacenters in a matter of minutes. Further, the converged approach Cato takes simplifies security and network visibility. Again, this is because the solution, in this case a full network security stack, is built into the cloud native infrastructure.

#3. Eliminate Packet Loss

Packet loss can wreak havoc on collaboration solutions such as VoIP and UCaaS (Unified Communications as a Service). Lost packets can be the difference between a productive business call or one where both ends become incoherent to one another.

The challenge in the underlying causes of packet loss can be anything from overworked routers to network congestion to software bugs. Cato’s cloud native infrastructure helps solve the packet loss problem using multiple built-in features including: business process QoS, dynamic path selection, active-active link usage, packet duplication, and fast packet recovery.

While roughly 1% packet loss can cause VoIP call issues under normal circumstances, RingCentral testing has shown Cato can deliver high-quality voice calls while experiencing more than 15% packet loss.

#4. Proxy TCP Connections

Fundamentally, TCP (Transmission Control Protocol) connections inherently add more overhead than their UDP (User Datagram Protocol) counterparts. At scale, this leads to scenarios where TCP connections can significantly contribute to network congestion and reduce throughput.

Cato PoPs help enterprises address this issue by proxying TCP connections to make clients and servers “think” they are closer together and allow for larger TCP windows. Further, TCP congestion control functionality enables optimization of bandwidth utilization.

#5. Aggregate Last Mile Connections

Blackouts and brownouts in the last mile of WAN connections continue to be one of the most difficult network performance challenges to solve. This is because the issues that can occur in the last mile and the infrastructure quality across the globe vary greatly. Aggregating last mile connections, ideally in an active/active configuration, allows enterprises to protect against the challenges of the last mile and improve network performance. Cato Cloud takes connection aggregation a step further and proactively monitors for both blackouts and brownouts and enables automatic failover when appropriate. Additionally, Policy-based Routing (PbR) helps ensure the optimum path is used every time.

Convergence is Key

Improving network performance given any particular network problem is one thing, but providing enterprise-grade connectivity at scale requires a holistic approach. This is where the converged approach of Cato’s SASE (Secure Access Service Edge) model shines. Optimizations and security features are inherently part of the network, simplifying deployment and management while also solving real-world network performance challenges. In fact, the simplified and streamlined approach is one of the things Yoni Cohen, CTO of CIAL Dun & Bradstreet, found most valuable about his Cato rollout: “I love what Cato is doing. They take an area that is complicated and make it easy,” says Cohen. “What we have done with them so far has made a meaningful impact on our ability to have a smooth transition to a unified company network and allowed this to be one thing that we’re not worried about.”
If you’d like to learn more about how Cato Cloud can help your enterprise, take a look at a demonstration or contact us today.

Dave Greenfield

Dave Greenfield

Dave Greenfield is a veteran of IT industry. He’s spent more than 20 years as an award-winning journalist and independent technology consultant. Today, he serves as a secure networking evangelist for Cato Networks.