How Amazon has become the black hole of the cloud world

By on

This article appeared in the June 2017 issue of CRN magazine.

Subscribe now

How Amazon has become the black hole of the cloud world

They’re called clouds, but they could just as well be called black holes. Amazon Web Services and those that have followed have demonstrated a voracious appetite for sucking in computing workloads, establishing economies of scale and a range of options that have knocked out competing solutions from many well-established competitors.

That hasn’t deterred many smaller local service providers from leaping into the fray, however, leading to a proliferation of cloud-like offerings across Australia. But in a market where size is proving to be everything, will the smaller providers find themselves crushed by the unrelenting gravity of the industry leaders?

In February of this year, US-based analyst firm Synergy Research Group estimated that public cloud services were now generating US$7 billion in revenue per quarter, at a growth rate of almost 50 percent. 

In the hierarchy of black holes, AWS has reached super-massive status, with an estimated 40 percent of the public cloud market. It dwarfs its next three nearest rivals in Microsoft, Google and IBM (which acquired SoftLayer for US$2 billion in 2013), all of which are massive in their own right, and as a group are catching up to AWS – albeit slowly. The trio made up 23 percent of the market in 2016: a full five percentage points more than the previous year, according to Synergy.

That leaves every other provider fighting over little more than a third of the market, a share that Synergy suggests shrunk by five percentage points in 2016.

Already, the harsh economics of public cloud computing have proven too much for several multinational providers, with HP, Cisco, Rackspace, Verizon and VMware all backing away from the market, and local telco Macquarie Telecom shuttering its Ninefold cloud service in 2015.

For Telsyte senior analyst Rodney Gedda, the difference between the leaders and the losers is commitment. “You can’t be half pregnant in the cloud,” says Gedda. “It will be the companies that absolutely commit to that business model that power ahead. 

“[Traditional providers] were trying to approach cloud in the same way as on-prem, so they stood up infrastructure on a build-it-and-they-will-come model, but charged people the same amount of money with a long onboarding process. And that just wasn’t going to cut it.”

Those that are in it for the long haul have been spending accordingly. IBM opened its Australian Softlayer data centres in 2014 and ‘15, while Microsoft Azure has had a local presence since 2014. Google will launch local hosting for its cloud platform imminently.

Can Australians compete?

The last two years has seen the exit of other established brands, with consequences for local partners.
ACT-based cloud management software maker Buttonwood Cloud Exchange, for instance, had to significantly alter its launch strategy thanks to Cisco’s decision to cancel its Intercloud service.

Cisco’s Intercloud fabric provided the data layer for Buttonwood to connect to AWS and Azure, which were its primary service providers, says Buttonwood’s managing director Allan King. He says that Cisco’s decision to step away from Intercloud “cost us 12 months, but what we did was what we were always going to do, which was to develop native integration into AWS and Azure. And we have just announced Oracle.”

King is not surprised that many larger players have exited the public cloud market.

“I felt they lacked the scalability required to provide the innovation needed by business,” King says.
“What they were really offering was a hosted private cloud or a VM-centric offering. Spinning up a new VM in another location is not that innovative. The need for R&D at the backend, to deliver what Azure and AWS are delivering, is phenomenal. And even organisations like VMware don’t have the deep pockets or the software layers to be able to deliver those offerings.”

That doesn’t bode well for many of the would-be cloud competitors in Australia, and King says asset refresh cycles might also prove insurmountable for many.

“You are competing with AWS and Azure, who are rolling out a billion dollars a year in new technology and infrastructure into data centres,” King says. “Every time they deploy an asset, they can leverage Moore’s Law to decrease price and become more competitive. If you are a small-to-medium organisation and you’re trying to sweat your assets, the only lever you have to compete is margin, and ultimately you can only pull that so far.”

For this reason, he suggests that many smaller providers may have a five-year lifespan.

“Generally, five years is the time you sweat an asset, and at the end of five years, you have to ask yourself whether you’ll reinvest or walk away,” King says. “And we are seeing a lot of those people walk away. I think the assets that have been purchased will be phased out and people will take on a more service-centric, managed service proposition.”

Not everyone is walking away though. Perth-based Zettagrid has cleared the five-year hurdle and established what managing director Nathan Harman describes as “a lovely little business”. He’s even been on the acquisition path himself, having identified around 200 different players in the Australian market.

“There will be lots of consolidation with the little guys who never quite got big enough and don’t have scale,” Harman says, “And we’ve been pretty actively looking.

“We’ve done one little acquisition, and we’ve got two in due diligence at the moment; our plan is to do at least one per year for a little while, and maybe even get slightly bigger as time goes on.

Harman says the factor that has enabled Zetta to survive was commitment to automation from the outset.

“What that allows you to do is lower your costs in terms of support and provisioning, and it also allows you to scale,” Harman says. “What we find is that businesses get to a size where they can only get bigger by adding more people or making a significant investment to add automation. And at that point, most stop. We are really reaping the reward from that now. With the acquisition that we did last year, Conexim, we were able to migrate entirely to our platform in very short order.”

The other key factor is ensuring that what’s on offer is unique. 

“Basically, what we do is take some best-of-breed software – Zerto is the last one we’ve done – and then provide it in a subscription model,” Harman says. “There is plenty of room for different niches. Some of those are more on the application side, while some are quite attached to managed services. It’s the pure-play guys, where if you can’t scale, you’re in trouble.”

For Sydney-based Hosted Network, the model has been to provide a white-label service that is on-sold by other providers. Managing director Ben Town says that when talking to MSPs and other partners, flexibility is key.

“That is where a lot of the smaller players really need assistance, in going beyond just a price book to help explain and architect it,” Town says. “And as a smaller organisation, we’ve got lower overheads.”

Opportunity abounds

Despite the apparent dominance of AWS and the shrinking of the remaining spoils, Gedda says there is still plenty of new opportunity for new players who want to enter. Alibaba, for instance, opened its Sydney cloud data centre in late 2016, while Oracle followed suit earlier this year. 

The market is big enough for diversity,” Gedda says. “You will always have oligopolies forming, but in terms of business IT service, the market is certainly big enough to support a top 20 of tier-two players. There is always going to be a need for an alternative. Do you really want all of your eggs in one basket?”

And there are more potential competitors still to arrive. Gedda points out that the second largest hosting provider in the world (as named by Netcraft) is New York-based DigitalOcean, a company that has raised US$123 million in funding and established facilities in Europe, Asia and India, but is yet to establish a presence in Australia. 

“They started selling Linux virtual private servers at a low price point, and they sell thousands of them,” Gedda says. “And they have a vibrant developer community.”

The crowded Australian market also recently witnessed the entry of European player OVH. That company shot to prominence in April 2017, when it acquired VMware’s vCloud Air facilities in the US and Europe. Gedda believes OVH has a reasonable shot at establishing a foothold in Australia.

“They are absolutely committed to that business model of cloud-only,” Gedda says. “They will build their own, and partner where it makes sense in data centre space, or even build their own data centres. And they will engineer their own stacks, servers, storage and networks and that sort of stuff. And for them, the economics of cloud make sense.”

OVH has signed up more than 1.1 million customers around the world, along with 5200 partners, and in May announced it would make Melbourne its regional headquarters. It will offer dedicated servers as well as private cloud and public cloud offerings.

According to OVH vice chairman Laurent Allard, the need for diversity is the key defining factor that has created an opening for his company.

“This is a fundamental point for our strategy, which is that one-size-fits-all does not work for cloud, as it does not work for a car or anything else in life,” Allard says. “We need diversity of technology, solutions and services, because we have diversity of needs. All of this diversity can be combined in a single virtual data centre controlled by the client, as what we call a vRack.”

OVH’s senior advisor for APAC, Emmanuel Goutallier, says the company will be looking to build out its partner model locally.

“We have already made a number of contacts and are in advanced discussions with a number of service providers and SIs,” Goutallier says. “And we see that our value proposition is quite sticky, because with our dedicated servers and private cloud, we give control of that environment to the SI or the service provider. We are not stepping over their control levels, because really, we’re selling a piece of dedicated infrastructure with the public cloud attributes.”  

An interview with Buttonwood Cloud Exchange

Well-known channel identity Allan King, founder of long-time ACT-based system integrator Infront Systems, made headlines in 2015 when he announced the launch of Buttonwood Cloud Exchange. The startup emerged after two years in stealth mode as a new way for customers to orchestrate, automate and manage workloads across Microsoft Azure and Amazon Web Services. 

Buttonwood was built on top of Cisco’s Intercloud fabric. This must have seemed like a sure bet in 2014, when the networking vendor was pumping a billion dollars into the Intercloud portfolio. Like King, top Australian Cisco partners queued up to join, with Telstra, Data#3, Ethan Group and Dimension Data all announced as early Intercloud partners.

But fast-forward to the end of 2016, and Cisco has pulled the pin on Intercloud. The decision left Buttonwood in the lurch and raised questions about whether anyone could compete with AWS and Azure. King talked to CRN about how Buttonwood pivoted to its new, post-Intercloud architecture, and the risks for partners in hitching their strategy to any one vendor in such a rapidly changing market.

What was the impact of Cisco’s decision to pull out of Intercloud?

It was a very costly exercise for us, and we learnt a big lesson along the way. As a startup, we chose Cisco because we were aligned with their strategy. We believed in their vision and felt we could complement this with our own investment and ideas.

Unfortunately, it proved that a billion dollars invested in this market is nowhere near enough to be able to compete with the hyperscale providers, even though Cisco was supporting AWS and Azure through their Intercloud fabric.

What was your background with Intercloud?

As a startup, we were very engaged with Cisco in San Jose though their Intercloud fabric development team. They had about 120 developers working on that. We made a deep, deep commitment to work them and help bring our joint vision to market. Unfortunately, two and a half years later they decided to pull the pin, which, as you can imagine, is quite disruptive for a self-funded startup.

How has Buttonwood moved ahead, given this setback?

Buttonwood was architected from the beginning to be a turn-key cloud management and integration platform, but initially, we built our control plane to support the work that Cisco was doing around Intercloud – the data plane. We felt it was a way to accelerate our go-to-market. We built the governance, integration and analytics to support Cisco efforts at the data fabric layer.

With Intercloud now gone, we needed to develop our own native integrations to support leading hyperscale providers AWS, Azure, Oracle and VMware. Our next integration is with the Nutanix AHV environment – one driven by customer demand. This required us to redirect a lot of our engineering investment to address that.

Now that we have completed the work and relaunched Buttonwood, we have a much more complete and mature product. We no longer have any dependencies on third-party solutions. This was something we were always going to do, but it forced us to bring these development efforts forward. It has cost us not only 12 months of delay to go to market, but millions of dollars in R&D.

What does this say about competition among the public cloud providers?

Hyperscale means hyperscale, and whether you are Cisco or VMware or a local provider, you can’t assume to compete on a global scale unless you have tens of billions to invest. Even though we were looking to Cisco to help accelerate our go-to-market, we only ever had plans to support the major providers – namely AWS and Azure. We learnt the hard way that even the biggest IT companies in the industry can’t compete at this scale.

I do believe that there is space in the market for niche providers, but not ones that peddle a simple VM replacement strategy. Organisations are looking to innovate, and this means significant value needs to be delivered above the line. Skills augmentation, full stack
service offerings, Tier-1 managed services... If you have a  simple IaaS strategy as a service provider, then I don’t believe they will survive.

What has been your takeaway from this saga?

I believe that there is a lot of opportunity in the market due to cloud and general technology disruption. Multinationals are looking to partners to help accelerate their vision.

I would caution only that you do your due diligence and structure any arrangement around your vision and not simply the vision of the vendor. When push comes to shove, even the biggest ‘partner-first’ companies will leave you holding the bag – often without a returned phone call when all is said and done.

How has this impacted your partnership with Cisco?

Cisco is a very impressive company but it shows that any business, regardless of size or reputation, can struggle to redefine themselves in an industry that is in a constant state of flux. As for us, we understood the risk. No business plan that challenges traditional norms is a straight line to success; there are always pits and traps and potholes. If it was easy, everyone would be doing it.

Would we have done anything different? Given the same information, probably not. However, I do believe we are now a stronger business for everything that we have learnt and endured. We not only have a superior product, but also a clear partnering strategy that is better aligned to deliver the outcomes in the world of new IT.  

Got a news tip for our journalists? Share it with us anonymously here.
Copyright © CRN Australia. All rights reserved.

Most Read Articles

Log In

  |  Forgot your password?