How many points of presence (PoPs) and what types of connectivity does your global network require to cover 95% of the knowledge workers with SLA-driven application performance?
1. How many PoPs do I need?
a. Less than 50
b. 50-100
c. More than 100
Correct answer is a.
Though you may think that more is better, what really counts is the PoP scalability and connectivity. 95% of the knowledge workers in the world can connect to only 30 PoPs in under 25ms, a low enough round-trip time to support sophisticated network and application optimization on the last mile. If someone tells you they have 100, or even 500 PoPs, ask them what type of hardware is deployed, how they scale, how they are interconnected, and how they are effectively orchestrated.
2. Which is better to connect the PoPs, Layer 2 or Layer 3?
a. Layer 2
b. Layer 3
c. Other
Correct answer is a.
Though much of the focus is on Layer 3 connectivity, the IP layer, and most SD-WAN solutions are built in this way, when looking at global connectivity, the IP layer is at the mercy of peering relationships that get in the way of being able to offer true end-to-end SLAs. Last-mile ISPs and first-mile cloud connectivity only adds to this uncertainty. A Layer 2 transport network cuts through all of this, delivering true end-to-end SLAs.
3. Where do I need optimization?
a. First-mile
b. Middle-mile
c. Last-mile
d. All of the above
Correct answer is d.
Many focus on optimizing the middle-mile, between PoPs, but in order to deliver true application performance, the SD-WAN service must also optimize the last-mile between the branch or HQ and the PoP, along with the first-mile connecting to cloud providers or data centers. Effective optimization requires a multi-segment approach where the different techniques are optimized for the first, middle, or last miles.
4. If I have sites across the globe, what connectivity is better for guaranteed application performance?
a. The internet
b. A dedicated Layer 2 transport network
Correct answer is b.
The internet isn’t capable of delivering consistent SLAs and application performance to highly distributed sites. In a single region where broadband internet may be highly evolved this may not be the case, but once you leave the region, a more deterministic means of connectivity is required. For some enterprises, a HybridWAN connectivity with a combination of dedicated Layer 2 for business-critical traffic and an internet option for cost-effective transport of non-priority traffic - especially within a region - can make sense.
Only the Aryaka core option offers guaranteed application performance around the globe.