SD-WAN in play: How software-defined solves core business challenges

SD-WAN to support global growth: HMSHost’s story

Part of the wider Italian Autogrill organisation, HMSHost is one of the world’s largest operators of food and drink outlets at airports, train stations and so on. It runs outlets at major airports including Amsterdam Schiphol, Beijing Capital, Copenhagen Kastrup, Dubai International and London Heathrow, and is one of the largest franchisees of Burger King and Starbucks locations in the world.

A separate US and Canadian business runs services at Boston Logan, Chicago O’Hare, JFK, Los Angeles, Seattle Tacoma, Toronto Pearson and Washington Dulles, among many others.

But with continuous growth of around 20% year-on-year, and new markets such as Australia coming on-stream, HMSHost was running into problems when it came to network management.

HMSHost’s Dennis Hoogreef, vice-president of IT and facilities, takes up the tale. “We see a combination of complexities,” he says. “In Europe, internet connectivity is relatively good, but outside Europe – and if you have centralised datacentres – latency becomes an issue.”

Latency is a particularly important issue for HMSHost because its point of sale (POS) platform relies on a backend datacentre in Frankfurt, Germany.

Speed of customer service is a crucial metric by which the firm is judged. Passengers in transit through airports are on a tight schedule and will get agitated if forced to wait around too long, so issues like a restaurant chip and PIN machine that won’t connect properly or takes a long time to work are felt more keenly than usual.

“If a plane is delayed, you could have 100 or 200 people walking into the cafe,” says Hoogreef. “Then you want to be able to provide efficient and quick service to them.”

Other complications

The same global spread that made network quality an issue also introduced other complications for HMSHost, such as the necessity to work with multiple connectivity suppliers.

“We had a couple of preferred [connectivity] partners on larger contracts, but as you know, companies get taken over or are unable to provide services everywhere, and we acquired ourselves a complex global spread which made it more diverse,” says Hoogreef.

“It’s not that we had a different provider in every country, but there were certainly multiple providers. We also often have to deal with landlords with specific infrastructure and specific providers.

“For us, Aryaka brought three things together. First, we had an MPLS [multiprotocol label switching] service with VPNs [virtual private networks], which was legacy technology and, for us, labour intensive to maintain, so we wanted to move to managed SD-WANs,” says Hoogreef.

“Second, in Asia and China we had latency and connectivity issues, so we looked at WAN optimisation. Third, we wanted all our connectivity to be monitored and proactively managed, including the last mile to each site.”

Strategic service provider

With Aryaka’s managed SD-WAN service ticking all of these boxes, and additionally offering private connections into HMSHost’s recent Microsoft Azure deployments, the team decided to move ahead, enlisting the supplier as a strategic service provider earlier in 2019.

“When you specialise in addressing the needs of travellers in a spot for a certain amount of time, you want to be in a position to guarantee their experience,” says Kiran. “MPLS was a complex construct for HMSHost to manage, and impeding its ability to be agile.

“Because network conditions are very different across different countries, it was important for HMSHost to normalise those and offer a consistent experience to travellers – regardless of location – without imposing a large overhead. That’s the reason they liked our infrastructure and went with it as an evolution of their MPLS offering.”

HMSHost can now operate over Aryaka’s 30 highly available, layer two meshed global points of presence (PoPs) on all six habitable continents to route credit card payment traffic as priority traffic at every one of its sites, improving the working experience for its frontline workforce, and service levels for its travelling customers.

It can now spin up additional bandwidth for expanding locations, and bring connectivity into new ones. It recently took on contracts to work on behalf of Chester’s Cheshire Oaks outlet mall and 30 Dutch railway stations – in days rather than months – without the hassle of working through its local providers.

It is also taking advantage of last mile management and low-latency guarantees through Aryaka’s managed services offering.

Looking ahead, HMSHost is now planning to leverage the global network from a disaster recovery perspective in the future. “We have a backup datacentre which can step in if it’s needed, but we still need to be able to redirect all connectivity to it, and of course now we have one global provider that should be easier to manage,” says Hoogreef.

SD-WAN in a divestment: Element’s story

HMSHost’s story  is typical of many who switch out MPLS for SD-WAN and realise benefits in network efficiency and business growth. But for Element Solutions Aryaka’s SD-WAN played a rather different role, helping the business through a recent sale of its agricultural solutions business, Arysta LifeScience to UPL, which completed in January 2019.

Just as SD-WAN can help a growing business, it can also help a shrinking or divesting one, but for understandable reasons this is a story that is told less often in the industry. “Element is actually a very interesting case study that puts the business context into how technology can either be an enabler or an impeding factor to making business transformation happen,” says Aryaka’s Kiran.

A relatively recent foundation, Element (formerly known as Platform Specialty Products) employs around 4,000 people in 50 countries, including the UK, and turned over $2bn in net revenues last year. Its specialist chemical products can be found in the components of electronic devices or vehicles, in the consumer packaging industry, and in offshore oil exploration.

Like many diverse businesses, Element grew through M&A and until the 2018 divestment functioned largely as a holding company for its various business units, something that is now changing.

Like HMSHost, it turned to Aryaka to solve the standard core infrastructure challenges arising from its growth spurt says CIO Dustin Collins; the need to implement a global network in a scenario where multiple lines of business were relying on multiple technologies and standards

Aryaka quickly proved its worth in terms of network agility, as Collins explains. “We are an asset- light business – we have a number of manufacturing facilities but we don’t have significant capital investment in any one facility. That allows us to change the supply chain to adapt to our needs, and there are significant network requirements derived out of that.

“Having something that’s extremely agile for us to be able to spin up or spin down in a matter of days – as opposed to months which is what you’d see from a traditional telco – is probably the number one reason why we’re with Aryaka,” says Collins.

Managed service style

Like his peers at HMSHost, Collins also calls out the supplier’s end-to-end, managed service style approach to the network, which he says differs from hardware-first players like Silver Peak or Riverbed, or network-first players like pretty much any telco you can name.

Ultimately, it all suits Element’s diverse, worldwide, and cloud-centric business nicely, so when it came to selling the agricultural business, Collins naturally turned to Aryaka straight away.

“Our engineering teams got together to try to figure out a way to do this in the least disruptive manner possible, because we’re continuing to operate and run the business,” says Collins.

Collins says it’s best to think of the Element network as being less of a standalone network, and more of “a tenant in the Aryaka network”, in the same way as other applications can be tenants of Microsoft Azure or AWS.

“In the least disruptive way, we tried to create two partitions with appropriate reliability and availability,” says Collins. “We needed to make sure the network ran and performed well, because the minute you start thinking about creating two partitions, you have to think about how to make sure the services on one side or the other still to operate and perform effectively while you’re splitting the network.”

Collins worked with Aryaka to build a bridge inside Microsoft Azure, using it as a central point of reference, with parallel virtual datacentres at each end of it which communicated across the bridge to each other, and to users.

“Once we had all of the datacentre pods split, we moved all of the services that were remaining with the divested company into their own pods, and then we shifted all the datacentre services for the divested company into their own Aryaka partition. Effectively, all of the users were still on the Element side, but accessing their services through the virtual datacentres across the bridge.”

Then, over a six-week rolling period, Element performed a series of cutovers moving site-by-site, 100 locations in all. Each switchover took about five minutes.

“Ultimately what we ended up with is the divested company running in its own Aryaka environment, Element running on its own Aryaka environment, and the bridge in place to catch anything that was still in the process of being transitioned, being carved out, moved, created multiple instances of, or whatever,” says Collins.

The advantage of this approach was twofold. First, it was less disruptive to users because the IT department could switch over very quickly during scheduled downtime.

Second, and critically, it avoided the need for complex and long-lived transitional services agreements (TSAs), meaning Element and the acquiring firm could cut the cord quickly and cleanly.

“It gave us the ability to flip a switch relative to the network in order to separate, so the very last thing we did was turn out the lights, and that gave us the ability to do that at one single choke point,” says Collins.

“The ‘gotcha’ moment – we didn’t realise how beneficial this was going to be – was no matter what kind of tools or capabilities you have out there to map your services, there will always be stuff that gets missed, or you didn’t know about.”

Aryaka report

But thanks to the Azure bridge and partitioned networks, Collins could simply have Aryaka generate a quick report to find out what services were still communicating across it.

“That gave us the ability to find services that weren’t fully documented or we didn’t have an understanding of, to be able to migrate or deal with them, avoiding a situation where you’ve done a network separation and then realised half the user community is screaming about something that’s got left behind which you didn’t know about,” he says.

“That was a huge advantage to us because it let us stage all of the separation and create a scenario where we cut the network on day two after close with no TSA.

“It speaks to the agility of the technology and the solution that Aryaka has created,” says Collins, who reckons it would have taken six months or more to split a more traditional network, and that’s before making both companies beholden to one another for far longer through TSAs.

“We set a goal of turning off the network on day one and so this almost forced the issue as it related to the entire organisation,” he says.

“For example, the financial group was forced to be able to have its own processes and procedures in place because there wasn’t this TSA sitting out there that everybody was using as an excuse for not being on time. It was a catalyst for us to stick to our commitments and accelerate the process. It helped make the break clean and positive on both sides.”

Cultural shift

For Kiran at Aryaka, Element’s successful divestment reflects to some extent the cultural shift currently taking place around networking.

“Besides SD-WAN, it’s also about the people,” he says. “As well as the Aryaka team, Element really came together as one team to make it happen. I’d give kudos to the intent and willingness of the tech leadership team on their side as much as I would to the underlying technology.

“Culture is always the hardest thing to overcome, and in some cases I feel cloud has paved the way for cultural shifts to happen in a more seamless way,” says Kiran.

“If you dial the clock back ten years, there was resistance to cloud and it was a mindset shift led by those who were comfortable enough to put workloads into it. That’s taken time for that transformation to happen and a lot of it was led by shadow IT.”

“Now I see the reverse happening; in most cases business users are comfortable with the cloud and so they do take a cloud first mentality, so if you’re doing something new, why not look at the economics and operational ease of the cloud first?

“That cultural shift to a cloud-first mentality has been driven top-down and aligned with digital transformation initiatives, so a lot of the tooling that’s now available has made it easier for technologies that are more fundamental and infrastructure-based to follow,” he says.

“The whole SD-WAN conversation would have been much harder to have had the cloud conversation not happened early on and paved the way.”