Get your quote
Blog post

The careful balance: performance vs cost

Technology choices matter. But cost choices matter too.

Ask the C-Suite what quality of connectivity they want, and it’ll answer “100%”. Always-up, always-on, for everyone, everywhere. They’re not concerned with fibre, copper, routing tables, or geographical differences. They just want the best.

But “best” is relative. It needs contrasting with other factors. Budget. Need. Business goals.

“Everything, everywhere” may be doable—but into-orbit expensive. Five-nines may be viable, and with less sticker shock. But here’s the thing: when you balance your needs with your budget, you may not need either. Or … you may need both.

Many businesses, even some large MNCs, work on 98% uptime and never even notice outages. While for others—typically, healthcare and banking—a minute of downtime in a month is the source of seriousquestions from the top floor.

It’s all a question of balance.

Now, cost/benefit analysis may not sound glamorous to a digital hero—until you recognise costs have always driven technological innovation. When a plough-plus-a-cow let one man do the work of twenty, that was a cost advantage. When the steam engine let a team of three transport tons—cost advantage. When undersea cables let countries communicate in seconds not weeks? Cost advantage.

That’s the subject of this article: understanding the cost/benefit balance in your organisation.

So let’s look at those choices now. With a bonus: findings from our research paper, “Understand Internet for SD-WAN and Cloud”. (To secure a copy of that report, click here.)

It’s in three parts: business requirements, cost model, and provider landscape.

Part 1: Start simple, with your business reqs

A great first step is to write down your basic requirements: the apps and data your people depend on. It’s harder than it looks.

Are you principally SD-WAN, hosting applications on your own servers? Have you joined the Cloud crowd, with thousands of people on GSuite or your backend on AWS? Or do you need quality across multiple sites, perhaps with a DIA pipe via a Tier-1 ISP?

In reality, you’re probably a mix. So make sure you know what proportions of each exist in your organisation. It’ll have an impact on your decisions later.

“Your connectivity reqs are defined by where your apps are and the routes your data take.”

Not all clouds look the same…

Example? If your apps are in the cloud, it’s not as simple as “broadband for everyone”. Cloud computing at enterprise scale—thousands of users—needs a critical eye on the paths data takes once it’s on the internet: your upstream routes.

The most seamless VPN or SD-WAN cannot compensate for an internet underlay with more holes than Swiss cheese.

For people in this position, DIA is an option. But a Tier-1 backbone connection (and the services you need with it) can be expensive … carrying guarantees you may not even need. In such cases, a high end fibre broadband may be the answer: faster than a DIA with scant seconds of downtime a year. A better match for your budget? Probably.

… while Shadow IT creates unseen problems

A flexible policy on software creates its own problems. 10,000 users running whatever apps they want means a lot of unplanned packets on your pipes. If those apps make them more productive? Great. But you’ve still got to provision for them.

(In some MNCs, shadow apps—IT in use outside the “approved list”—are the bulk of bandwidth. Do some CIOs think they’re running datacentres, when all the action’s actually in the Cloud?)

So first ascertain where your apps live and what routes connect them. Not the ideal case from your departmental strategy—but what people actually use. Because that’ll answer the basic question of how you connect.

Part 2: Take it up a level, with the cost model for each

If it’s 3AM and you know where your apps are, the second Big Question is to find whether the costs really justify the financial outlay. In a surprising number of MNCs, they don’t.

Over-specifying is a bigger issue than many in the C-Suite realise.

The CEO may want six-nines uptime. But what if that seconds-a-day outage for backups could be managed in the dark hours, inconveniencing nobody? 99.99 or 99.9% may provide a far friendlier cost case.

What matters to users, matters to you

Customer satisfaction (CSAT) may seem a long way from technical issues like bandwidth or network policies. But a scheduled backup at 7AM instead of 2AM may have a huge impact on an early riser’s productivity. And if you’re serving 200,000 users, even minutes of downtime make a difference.

One leading automotive group stands to lose millions if there’s an hour or two of outage. Yet another company with a similar architecture may suffer no continuity snags at all, simply because the issue doesn’t affect their daily productivity. People have different satisfaction criteria. Always ask yours what they think is important.

Diversity of connections, or not?

Access resilience is all about backup. So most MNCs specify multiple connections to each site, often with a different provider. Even physical routings are second-sourced, providing an alternative path for data that avoids chokepoints in Hong Kong or New York. Costs here take on an extra dimension: the cost of not having them. You need to consider these costs, too.

“What are the consequential losses to your business of an hour’s downtime in a vital location? Make sure these estimates go into your model, too.”

Beware the single-site cost spike

Middle-mile solutions can make business broadband feel like high-end SLA-backed services, providing a customer experience as good as you’d get from the big carriers—at a fraction of the cost. Conversely, if your SD-WAN links datacentres in many far-flung sites, a single one in the Middle East (where DIA connections are pricey) might upset your entire cost structure worldwide.

So your cost model is about more than bandwidth or PoPs. Fortunately, there are answers.

“Cost/benefit analysis is all in the details.”

The lesson here: look critically at where it takes big resources to meet small goals. That’s where high costs hide.

Part 3: Get the big picture, by looking at the landscape

The third factor involves the global connectivity landscape. With all its twists and turns at the local level. It comes down to access type availability.

Here, there, not everywhere

Business broadband may be cheaper. But it rides on the back of consumer networks designed to serve households. Which means the fat pipe that gives jitter-free Netflix to a suburb may not reach your rural office park.

And even when you’ve factored in differences between countries, there’s still variance within countries. Kenya and Vietnam have brilliant 4G … but good luck getting a fibre spur to your industrial site.

Beware! Resource hogs ahead

Also look at utilisation and optimisation. Because without predictable routing, packet loss and latency may be eating both your bandwidth and your people’s productivity. Again, your upstream routing matters here.

The box the service comes in

Last but never least is the SLA that wraps your services. Riffing on Part 2, providing that five-nines 1Gbit service level across all your sites may be very cheap in some locations … and verypricey in others. (Hi, Mongolia!) And when things go south, the break-service contract comes to the fore. What if your business requirement is a 30min response time, but providing that human voice in central Africa adds €100,000 to the SLA?

Stress again: such outliers are often hidden. The world’s diversity is wonderful—but that’s not helpful if it means you have to manage hundreds of ISPs each with different SLAs.

Of course, there are other factors. (Many detailed further in “Understand Internet for SD-WAN and Cloud”.) But these three loom largest: business requirements, cost model, and connectivity landscape. Consider all three, and you’re well on your way to achieving the optimal balance between cost and return.

Conclusion: keep the questions coming

Optimising SD-WAN. Rolling out Cloud. Tier-1 DIA. The mix is up to you. But many global organisations are stirring another option into that mix: ISP aggregation.

With high-quality business internet from ISPs close to your sites worldwide but covered by consistent service level agreements everywhere, you get the assurance of local expertise, without the complications of dealing with a hundred local ISPs eating into your internal resources. (Your people have better things to do, after all.)

And best of all, it’s not an all-or-nothing choice. ISP aggregation can fill in gaps in your service map, flatten those cost spikes by being able to match local knowledge with the right access technologies, and smooth out local conditions—without having to rip-and- replace your existing infrastructure.

Every solution has its merits. And ISP aggregation is proving the right balance for over 600 Globalinternet customers. Maybe it’s the right balance for you, too.