Top Of The SASE PoPs

In my previous blog I noted how themes occur seemingly from nowhere, that are common among vendors from different spheres of IT.

In this particular case it is various vendors I’m working with who have been helping customers with problems created by under-performing service delivery, all traced to a lack of local PoP/supplier network. Yes – it is that fundamental. Imagine a SaaS platform where the nearest PoP isn’t even in the same continent; that’s the reality we are talking about here. Another prime example of a platform that relies on optimised, consistent performance is the much aforementioned SASE – delivering networking and security services from a cloud-based network of PoPs and sub-PoPs. So, putting those PoPs next to customers – or providing the optimisation that delivers the equivalent results – is critical for the same reason it was critical to Aritari’s VoIP customers (in my previous blog), notably in order to reduce latency. Too much latency bringing traffic to a PoP for processing will undermine the user experience, meaning classic issues of lower productivity, higher support costs and – in some cases – simply an inability to run certain, possibly critical, applications.

From an end user perspective, the question is: how do I effectively evaluate a potential SASE supplier from the PoP perspective – how many and where are they geo-located, especially in respect of where use user base lies? As ever within the world of IT – i.e., the biggest marketing game in the world – vendors seem to have different claims in terms of PoP count, or even the basics of what their delivery networks actually consist of, full stop. It really is a case of differentiating the marketing “box tickers” from the genuine mass deployment examples. Even more complex is the issue of what is being defined as a PoP; for example, a basic network node for ingress traffic is not a PoP – it simply directs backhaul traffic to genuine compute PoPs for processing, which can thereby impact on performance again. And just how many functions should a PoP have? Are all PoPs created equally? No, they are not.

Given that SASE/SSE is one of those classic “get on the bandwagon” Gartner inventions, it’s important to be able to differentiate between those who are serious, dedicated players, and those who have simply said “I think we can do that as well”. The network underpinning the service, then, is key. And working out who has what and where is indeed confusing. From vendor to vendor, the available public info varies significantly, but there is something to work on there, in most cases at least.   Think of this as an initial triage process for creating a shortlist of vendors to evaluate in more detail. Obviously, without putting any service to the test, there is no “absolute” in this approach, simply a means towards an end. And, of course, online data can also be out of date: yes, that is a failure on behalf of the vendor, but it does mean that what you read today is not actually the reality of tomorrow, but there has to be a starting point somewhere.

What it all boils down to is that there is always far more complexity in selecting a vendor-provided service than you would initially anticipate – or want. But then, no one said IT was easy – otherwise we’d all be out of a job!

Data Center
Data Management