Today, there is a lot of effort toward speeding up cloud-hosted app performance through the use of solutions like SD-WAN, but not enough attention afforded to the so-called last mile of these connections. This last mile is the final network step that gets data coming into a particular office to the user requesting that data. For IT teams managing any type of remote location and the WAN that goes with it, the last mile is essential.
Beyond the last mile lurks another final step that is as essential to performance: corporate WiFi. The WiFi network has cemented its place as a business-critical part of many modern enterprises. Here are some considerations in figuring out how to make last-mile and last-hop performance better.
Choosing MPLS vs. Broadband for the Last Mile
Will you choose MPLS or broadband for last-mile WiFi at remote offices? For the near future this will still be a debate, but it’s a foregone conclusion that broadband can serve many companies well as a corporate solution. Technologies like SD-WAN will ensure that broadband is up to the task. There will, for the time being, remain a number of instances where the SLA-based reliability of MPLS is the best solution. The number is falling rapidly as broadband speeds offer more flexibility and “good enough” security at a lower cost.
For some industries, this transition is a bit more complex and the use of hub and spoke architectures through cloud security proxies will still reign supreme. However, with VPN tunnels and traffic segregation, there are more ways than ever to ensure secure delivery of packets across the WAN.
Managing BYOD Traffic
The bring-your-own-device trend is still in the news, but for many has transitioned from a problem of strategy to a problem of execution. Networks that were originally separated at the WiFi access level now must be separated by circuit or QoS marking. That ensures that those BYO devices don’t congest the links that support Office 365, G Suite and other crucial apps in use today.
Another common tactic is to ensure that recreational traffic doesn’t negatively affect business- critical applications by using split tunneling to relegate recreational traffic to the public internet. Then, MPLS can be used for those important apps. However, even these configurations need to be monitored. It is the network or IT ops groups that must keep a watchful eye through proactive alerting. If one of a pair of broadband circuits has a disruption as part of an SD-WAN deployment, then traffic is back to a single point of failure, or more likely, congestion. MPLS is famous for having better SLAs than broadband, but you sacrifice visibility.
Using QoS for the Last Mile
Typically, the last mile refers to the ISP handoff to a regional provider, where QoS is extremely relevant if you are paying for it. QoS is a metric that needs to be honored from end to end in order to provide any benefit. Monitoring QoS is therefore essential. A single hop that does not honor the DSCP markings of QoS will likely result in “Best Effort” and remove any performance benefits.
Tracking Bandwidth vs. Capacity
Broadband providers sell bandwidth, but good user experience depends on capacity available from the user out to the application. Bandwidth is a point-in-space measurement of a device’s ability, limited or not, to transfer data. Capacity is the achievable transfer speed over the entire application delivery path, from user to application. The FCC says that the speed of broadband should be greater than or equal to 25 Mbps. Traditional tools to determine speed require flooding the network. Now, though, monitoring tools are starting to come around to a continuous approach that can accurately profile changes to the capacity of the full network path, whatever critical apps and endpoints are included.
Isolating the Last Hop in Practice
AppNeta uses a unique approach to identify performance of the last hop of the network. A Small Office Monitoring Point at a remote location can join the WiFi network as a member, using the same connection as your users. Our customers can then configure network tests between an office location and a SaaS or web app using two network paths. One path is generated from our device behind the firewall near the ISP egress point and directly accesses the internet connection. Another test uses the WiFi to connect and gain access to the same connection. The difference between the latency of these two paths shows the WiFi connection. To add even more end-user realism to the test, customers can configure periodic teardown of the connection to simulate the actual time for users to establish an IP through DHCP and then connect to an external point.
Keeping an eye on last-mile and last-hop network performance will pay off with better application performance and happier users. IT network teams actually can see end-to-end in today’s networks.