For better or worse, multiple clouds are here to stay
In information technology circles, it’s called “one throat to choke.”
It’s a metaphor for chief information officers’ preference for concentrating most of their business with a single strategic supplier in each category of application and infrastructure. The approach has a lot of appeal to risk-averse IT organizations, including fewer points of failure, better customer service, bigger discounts and clearer strategic direction.
It also gives birth to superpower companies that come to dominate their markets by virtue of being the safe choice. Giants such as Cisco Systems Inc. in networking, Oracle Corp. in database, Microsoft Corp. in desktop software and EMC Corp. in storage all rose out of the brutal scrum that characterizes the early years of nearly every emerging technology market.
“The leader makes money, number two does ok and everyone else is lucky to break even and – well it’s not even worth staying in business,” said David Vellante, chief analyst at SiliconANGLE sister company Wikibon. For the most part, CIOs are fine with that. Dominant suppliers simplify decisions and carve paths that others follow.
But when it comes to cloud computing infrastructure, that pattern hasn’t repeated itself. The biggest platform providers – including Amazon Web Services Inc., Microsoft Corp., Alibaba Group Holding Ltd. and Google LLC — are getting bigger, and none is a shoo-in for CIOs’ business, and certainly not on an exclusive basis.
The result, most experts agree, will be a multicloud world – and all the complexity that comes with it – for a long time to come.
That may frustrate some IT leaders who value simple choices. “CIOs like the ability to arbitrage, but when it comes to building apps we see them preferring a single cloud,” said Milin Desai, general manager of cloud services at VMware Inc.
But the reality is that “a single cloud, much like communism, sounds great in theory but isn’t practical when applied,” said Doug Saunders, CIO at Advanced Disposal Inc.
SiliconANGLE polled about two dozen industry executives, analysts and CIOs on the question of whether today’s complex multicloud landscape is a pit stop on the road to a more homogeneous future or a long-term reality. The consensus: “Every indication is that [multiple public clouds] will be the future,” said Ed Anderson, vice president and distinguished analyst at Gartner Inc. “Putting all your apps in a single cloud environment is an unacceptable risk.”
The issue will take center stage in San Francisco this week at VMworld, VMware’s annual conference. VMware is one of several large technology providers that, having failed to make a dent with its own public cloud offering, have pivoted to helping customers manage the multiple public and hybrid clouds that are the reality at many large companies.
Rather than waiting for clear choices to emerge, enterprises should learn to live with diversity, experts agree. While dealing with a patchwork of platforms, services and interfaces that are proprietary and often incompatible creates administrative overhead, the pain is worth the gain if organizations take advantage of the leverage and choice that competition gives them.
“Any enterprise that isn’t hedging its bets [by leveraging multiple cloud providers] is missing out,” said Scott Crowder, CIO at BMC Software Inc. “You need a credible threat if you want to negotiate.”
That thinking may fly in the face of the “one throat to choke” rule, but the cloud ecosystem is so vast and the range of services so diverse that it’s unlikely that one provider will ever be able to do it all.
“The public clouds all have flaws,” said Brian Gracely, senior director of product strategy for the OpenShift product line at IBM subsidiary Red Hat, citing recent Gartner research. “AWS is doing a poor job integrating all its new services. Microsoft Azure continues to have outages. Google is still trying to figure out how to sell to the enterprise.”
Some cloud providers beg to differ. Amazon, which is currently the undisputed king of the hill, believes that using multiple cloud platforms forces customers to standardize on the lowest common denominator, duplicate administrative effort and forfeit attractive discounts. The company’s stated position is that most customers will ultimately pick one predominant provider. The cloud giant’s branding guidelines even prohibit references to the existence of multiple clouds or alternative providers.
But there’s plenty of money riding on an alternative outcome that will see customers spreading their investments around and shifting workloads between platforms to take advantage of the best deals and technologies.
“We want to give customers the option of running their workloads in the environment best suited to their needs,” said Jennifer Lin, director of product management for Google’s Anthos platform. Announced last spring, Anthos is a bold attempt by Google to take advantage of the multicloud environment by giving customers a set of open platforms and development tools that enable them to easily shift workloads.
Sport of kings
The cloud infrastructure market never went through winnowing-out the process that has characterized many new technology platforms in the past in large part because competition from below was never a factor. Startups are typically the disruptors in new technology markets, but the massive infrastructure requirements of public cloud have dictated that only the largest companies could play. “If you were going to try to build a public cloud in western world, I can’t imagine it happening today,” said Mike Tuchen, chief executive officer of Talend SA. “It requires billions of dollars of investment every quarter.”
Cloud computing also isn’t a platform in same mold as operating systems and network protocols. Infrastructure-as-a-service providers offer a vast variety of products – more than 165 for AWS alone – giving customers a bounty of choice. “The cloud is big enough and broad enough that it doesn’t behave as we’ve seen platform markets behave in the past,” Tuchen said.
That consequence of such a fragmented landscape is that customers have to pick and choose from a bewildering array of options, which can be a problem or an opportunity, depending on your perspective. Amazon is the undisputed leader with 48% of the IaaS market, according to Gartner. However, Microsoft, Google and Alibaba are all growing at a faster rate. Regional and specialty players also account for nearly one-quarter of the market.
Although the various cloud platforms all offer the same choice of operating systems and basic services, “There’s a misnomer that cloud is commoditizing,” said Gartner’s Anderson. “Even if that’s happening it’s at the lowest level of the stack,”
That’s a good thing for customers, said Craig Lowery, a research director in the technology and service provider group at Gartner. “If all cloud providers were all the same you wouldn’t see this level of innovation,” he said. “There’s value in using more than one.”
In fact, cloud providers have little incentive to enable easy interoperability. Their applications and APIs are mostly proprietary, and all make it expensive and inconvenient to move data off of their platforms. “We’re seeing more diversity rather than less,” said Talend’s Tuchen.
While all providers acknowledge that customers want some degree of workload portability, making that process too easy is an existential risk. “If cloud looks like health insurance, customers will treat it like that,” said Rich Petersen, co-founder of JetStream Software Inc., referring to the common perception that one insurance plan is pretty much like any other.
How did we get here?
The current multicloud environment evolved through a combination of happenstance and necessity.
In the early days of cloud infrastructure, many organizations gave business users responsibility for selecting their own cloud platforms. Some of those applications have since become critical to the business and shifting them is now impractical. “Multicloud is largely a symptom of multi-vendor,” said Wikibon’s Vellante. “It’s companies acquiring clouds for specific workloads, shadow IT and pockets of activity versus a strategy of managing across multiple clouds.”
Those platforms are now so entrenched that many organizations have little choice but to live with them. “The typical multinational has a bunch of different divisions and each has its own cloud,” said Corey Quinn, cloud economist at consultancy The Duckbill Group. “Do you turn them off? Not unless you enjoy pain.”
Mergers and acquisitions have further complicated the picture by bringing new cloud platforms into the IT mix, presenting an entrenched integration problem for fast-growing companies, in particular. Some regulatory demands, particularly in Europe, require data and applications to be hosted within certain geographical boundaries, forcing IT organizations to spread workloads across an assortment of regional vendors.
And then there’s data gravity, a term for the cost and difficulty of shifting large amounts of data from one platform to another. The egress fees that cloud providers charge can make the cost of shifting data off of their platforms more expensive than keeping it there, turning their services into what JetStream’s Petersen called a “roach motel” in which clients check in but never check out. “Data tends to be sticky,” said Gartner’s Lowery. “It forces you to cling to a cloud.”
CIOs who desire workload portability are learning to cope and even prosper by keeping workloads logically separate and playing vendors off against each other. “The most successful organizations will find a way to embrace multicloud,” said Brian Kelly, chief executive of CloudBolt Software Inc.
A host of technology providers are rushing to help. They include VMware, IBM with Red Hat and Cisco, three giants that failed to crack the top tier of public cloud providers on their own and so instead have focused on helping customers federate multiple clouds.
“We are intending for the client to have a seamless, automated experience across environments,” said Hillery Hunter, chief technology officer of IBM Cloud. “What lies underneath is a choice of convenience, geographical boundaries, features and functions.”
These companies are hoping to be the brokers for their customers’ cloud operations, and perhaps de-position the platform providers a bit in the process. But the jury is still out on whether a complex technology environment is simplified by overlaying more technology on top. “Do you need another cloud vendor? Probably not,” said Quinn.
A growing number of systems integrators are recasting themselves as “catalog navigators who are dedicated to understanding what options are available and can do it for hundreds of different customers,” said JetStream’s Petersen. The pitch is that “you can have a multicloud strategy without hiring 20 or 40 more people,” he said. For example, VMware’s Cloud Provider Partner Program now encompasses more than 4,300 partners that are building on its cross-cloud management products.
Plan for the platform
IT organizations that allocate workloads to the most appropriate platform can take advantage of the strengths of individual cloud providers – such as Amazon’s breadth of services, Google’s machine learning expertise and Microsoft’s collaboration software estate – without creating management headaches. While casting lots with a single provider may yield some enticing discounts, the benefits of arbitraging between competing services is often even better, CIOs said.
Advanced Disposal’s Saunders believes that concentrating on a single cloud platform for the sake of reducing risk is misguided. “More risk reduction can be found in adhering to open standards and reducing ownership of your own infrastructure where possible,” he said.
That’s the approach Thomas DeChiaro is taking. The CIO of Drexel University in Philadelphia uses standardized software, avoids data-sharing across cloud platforms and dictates the terms of engagement for providers that want his business. “Even though platform providers want to push you into their own product suites, Microsoft is not going to tell me I can’t use Oracle,” he said.
Drexel runs most of its IT workloads on premises, in part because of the superior economics of the shared-service model the university employs. It primarily uses cloud providers for backup and disaster recovery purposes so that it can recover to its own infrastructure or another cloud. And it keeps things simple. “Spinning up an instance is pretty much the same from one provider to another,” DeChiaro said.
The benefit of that approach is “more flexibility without that much more complexity,” he said. “If I have workloads in two cloud providers and one raises prices, I have flexibility with another.”
Not everyone agrees that the price leverage of using multiple clouds is worth the overhead. “The notion of cloud arbitrage is largely fiction,” said Michael Bushong, vice president of enterprise and cloud marketing at Juniper Networks Inc. “Any cost benefits are offset at least partially by the difficulty of creating portable applications.”
Which is largely a function of people. Multicloud environments carry with them a skills premium, since the expertise needed to administer one cloud isn’t necessarily portable to the others. Autodesk Inc. dedicates a staff of 80 IT professionals to managing its AWS operations; adding Microsoft Azure to the mix would require hiring 40 more, said Sam Ramji, vice president of cloud platform at the software company Autodesk Inc., said in an interview earlier this year.
That’s one reason Google says its multicloud approach is superior: By adhering to a single platform, customers can shift between clouds without having to hire more developers and administrators. “The cost equation today isn’t in license costs,” said Google’s Lin. “It’s in operating expenses and retraining.”
Google’s Anthos strategy is rooted in Kubernetes, the popular open-source orchestration manager for the portable software environments called containers. Although Kubernetes solves part of the portability problem, it only does so if developers don’t reach outside to tap an individual provider’s proprietary services.
“Kubernetes is overly complicated, requires a reimagining of how systems live and isn’t simple to maintain,” said Duckbill Group’s Quinn.
Added VMware’s Desai: “Kubernetes helps accelerate innovation, but it doesn’t solve the complexity problem.”
Are cross-cloud apps the future?
One area of universal agreement is that the Holy Grail of applications that federate multiple clouds and borrow the best from each is a long way off for reasons that include complexity management and security. “If you’re planning to build something greenfield that lives in multiple cloud providers it’s almost always a bad idea,” declared Quinn.
Even Google, which has made multicloud coexistence the cornerstone of its strategy, doesn’t see demand for cross-cloud applications anytime soon. “We are focused on giving customers the ability to write once, run anywhere,” Lin said.But the first organizations that tap the potential of distributed cloud applications may change entire industries, said John Seely Brown, the former Xerox Corp. research chief who was on the forefront of many of the industry great innovations of the 1970s and 1980s.
Brown believes innovations such as blockchain will combine with cloud to reshape the way commerce is done. New kinds of enterprise resource planning systems will be built that span multiple partners in a supply chain with smart contracts and permission blockchain ledgers stripping away administrative overhead while ensuring security, audits and data integrity. Data gravity won’t be a problem, Brown maintained, if data is spread across multiple locations and federated via blockchain.
“I can compose things like Lego blocks from services as opposed to force-fitting the way I want to do something into an existing system,” he said. In short, the multicloud will reinvent the way applications are built.
Brown said most organizations will struggle initially to wrap their minds around the idea of building software that changes according to the needs of the business rather than forcing conformity to code as has been the case for the past 50 years. But a new platform presents an opportunity to rethink assumptions.
“If we use the same tools we used in the 20th century, the possibilities of building radical new businesses will be constrained,” Brown said. “The real value is in reinventing institutions.”
Few cloud adopters are thinking much about reinvention at this point, though. They’re trying to wrestle a multiheaded cloud monster to the ground with the conviction that today’s hard work will lead to a simpler, easier and more flexible future. Even if that means having to choke a few more throats in the meantime.
A message from John Furrier, co-founder of SiliconANGLE:
Your vote of support is important to us and it helps us keep the content FREE.
One-click below supports our mission to provide free, deep and relevant content.
Join the community that includes more than 15,000 #CubeAlumni experts, including Amazon.com CEO Andy Jassy, Dell Technologies founder and CEO Michael Dell, Intel CEO Pat Gelsinger and many more luminaries and experts.