On the morning of February 28, 2017, Amazon Web Services (AWS) experienced an outage in the US-EAST-1 Region among systems relying on its Simple Storage Service (S3) cloud offering. The event wreaked havoc on thousands of websites. The cause? A fat-finger mistake by a human in a line of code that took more servers offline than desired.

“Removing a significant portion of the capacity caused each of these systems to require a full restart,” noted Amazon in a blog post. “While these subsystems were being restarted, S3 was unable to service requests.”

A few weeks later, on March 21, Microsoft’s Azure cloud suffered an outage, knocking US East Coast users of Office 365, Outlook.com, and Skype—among other services—offline. As Waters went to press, the cause had not been released.

Amazon and Microsoft represent the two largest cloud providers, globally. According to Synergy Research Group, in the public cloud services space through the fourth quarter of 2016, AWS holds the largest share—just over 40 percent—though Microsoft (Azure), Google (Google Cloud Platform, GCP) and IBM (IBM Cloud), are all gaining ground. The next 10 providers have lost market share compared to these four. Synergy estimates that the public cloud infrastructure space—including both infrastructure-as-a-service (IaaS) and platform-as-a-service (PaaS) offerings—is “well over $7 billion and continues to grow at almost 50 percent per year.” 

Amazon’s cloud offering is dominant in the capital markets sector, but there’s a sea change afoot. In the years to come, vendors that service both sell-side and buy-side institutions will likely expand the way they interact with cloud providers, preferring to employ a multi-cloud strategy. And banks, hedge funds, exchanges and asset managers will be more willing to turn their data over to these companies directly—something that was thought to be preposterous just five short years ago. 

While Amazon (especially) and Microsoft (to a lesser extent) have jumped out on the others, Google is making a major push in the capital markets space and is gaining ground rapidly, announcing some major client wins in recent months. IBM has been making a push in the blockchain space and could make its presence better known if it can commoditize its Watson cognitive services on Wall Street to greater effect. Even technology giant Oracle is looking to throw its hat into the ring, with its executive chairman, Larry Ellison, saying last September that the company’s main objective was to cut down the AWS lead. 

This also doesn’t mean that firms are leaving AWS—at least not in the capital markets. In fact, during a panel at this year’s North American Trading Architecture Summit, which was held on April 5, executives from Axa, Credit Suisse, New York Life Insurance Company and TD Bank, all said that they are using both AWS as well as another of the major public cloud providers. This points to a blending of public cloud providers on Wall Street. While there will always be roadblocks for these behemoths—sensitivity of data will still detract many; foreign laws will prevent others—their influence will only grow as they look to chip away at AWS. It’s a story of coexistence, not replacement. 

Added Flexibility 

For this feature, Waters spoke with over two dozen industry experts about the reasons for AWS’ dominance, and where other providers might find headway. While price will always be a major factor when it comes to picking a cloud provider, price improvements are what’s expected. It’s the added tools, functionality, geographical expansion, security resources and task distribution that carry the most weight. And right now, it would appear that Google is finding the most success. 

Symphony Communication Services is one of those vendors that have already made the move to the bifurcated cloud system. The Palo Alto, Calif.-based communication and collaboration platform provider launched at the beginning of 2014 using AWS as its back-end cloud provider. In July 2016, it announced that it was also going to connect to GCP, allowing customers to choose between the two as their default storage provider. (It should be noted that Google is an investor in Symphony.)

The company’s founder and CEO, David Gurle, says Symphony’s architecture is reliant on microservices, which depend on the computing power and high-end operating system services of cloud providers.

“From an engineering approach it really made sense to be able to service a wider variety of clients and to take the best of each available service,” he says. “The fast connectivity between the different systems really enables us to build a service that can span two—maybe even three—cloud service providers.”

A great benefit of bifurcation is that by using both Amazon and Google, should there been an outage or jarring latency spike, Symphony’s services—because they’ve been architected to both—can easily roll over to either platform. 

Then there are those that are considering adding Google to their toolkit. London-based CloudMargin, which specializes in collateral management workflow solutions, is currently an AWS user. Steve Husk, CloudMargin’s CEO, says, though, that a multi-cloud solution would allow for flexibility so that two or three years later, should the market move, they can more easily adjust. 

“To maintain a level of independence, I don’t want us to be tied to AWS; AWS is a fine platform and it provides us with a continuous, scalable level of performance for any number of services, but we have foreign clients who may not want to use AWS; AWS would be a no-go area for them.”

He also says that while AWS has a lot of “very nice tools,” such as its cloud monitoring service CloudWatch, depending too much on those services “would tie us into the platform,” he says. 

Lee McCormack, head of strategy and product development for the vendor, says that this isn’t a life or death struggle. “The market is big enough for multiple players—Amazon, Google, Microsoft, Oracle—this is not a winner-takes-all,” he says. “Amazon is way ahead but there’s room for others; the market is going to grow.”

Cool New Tools

Trading Technologies (TT) has its two-year-old software-as-a-service (SaaS) delivered TT trading platform on AWS. It tapped AWS because of its range of services, such as CloudWatch and CloudFormation, which allows users to set up stacks of AWS resources in a “rubber-stamp fashion,” and the Cassandra cluster offering, which TT uses to store all encrypted data. 

But TT’s CEO, Rick Lane, says the platform will likely be bifurcated between Amazon and Google in the near future.

“What we find ourselves doing—and we’re trying to get better at this—is we’re now taking a lot of those things that we’re building ourselves and seeing if Amazon or Google—from our perspective, it’s a two-horse race—offer a platform that gives us this stuff out of the box where we don’t need to manage deployment to servers or manage scaling out when read or write throughput needs to change,” Lane says. “What we’re trying to embed in our product development team’s mind is, when we’re getting ready to build something, let’s ask ourselves if Amazon or Google offers that as a service.”

A game-changer for Google, according to Lane, could be its Cloud Spanner offering, which is its globally distributed relational database service. In February, Google launched the public beta version of the technology. Spanner would represent the first relational database that has global consistency and the ability to write data to it anywhere around the globe. 

“Right now, a lot of the relational databases that we have, there’s one region where we can write to—such as US-EAST-1 for Amazon—but we can read from anywhere. That’s good in some cases, but if we have traders in India or Singapore or Tokyo, we want their writes to be fast, as well. We don’t want them to have to come back to Virginia,” Lane says. “Spanner could fundamentally change the way that we do general relational databases. The benefits that it will provide us will be too great for us to care about having two cloud platforms. So we will be bridging the two, almost certainly.”

Then there are those that are both expanding their AWS footprint while considering the addition of another cloud-service provider. For example, LiquidityBook, a trading solutions vendor based in New York, is expanding its usage of AWS. It’s planning to leverage AWS’ IaaS platform to allow it to expand in markets where LiquidityBook currently doesn’t have datacenters, and it’s currently migrating its core operations to AWS.

“We’re moving our core operations to AWS for a few reasons,” says Shawn Samuel, LiquidityBook’s CTO. “The cloud lets you expand globally in a way that you can’t if you’re trying to open co-located datacenters if you’re a small firm. We can stand up overnight in the EU and Asia with AWS and that’s something we couldn’t do with the physical footprint we had in the US. Another big reason for the move to go entirely cloud versus a hybrid model is the operational scalability it provides. We’re a fast-growing firm and while we have a nice footprint in the US, the ability to scale 10 times or 100 times with the flip of a switch—that’s on-demand computing access.”

But, that doesn’t mean that the company is locked into AWS.

“Even though we’re not going to implement another cloud provider initially, we’re trying to stay multi-cloud,” to future-proof the company, Samuel adds. 

Cracks in the Wall

AWS is the undisputed leader in the space. It’s been able to lower its prices consistently while adding new tools, such as Amazon DynamoDB, its NoSQL cloud database service; Amazon Redshift, a data analytics offering; and AWS Lambda, a server-less compute engine that allows users to run code without provisioning or managing servers. The provider launches a staggering number of services per year—in the range of a thousand-plus. 

Others, though, have caught on to this model and better understand the necessity of it today than perhaps three years ago. 

After the AWS S3 outage in February, a bank CTO whose firm uses S3 to power its customer relationship management (CRM) platform, told Waters, “We’re just sitting around doing nothing.” The institution couples AWS with Azure, tapping into Microsoft’s Blob service for disaster recovery. The bank is looking at Google and IBM to see if they can help it cut down on its processing needs to run deep analytics.

A CIO of a hedge fund that deploys machine-learning technologies within the firm is looking at IBM to run alongside its AWS hookup. “They’re all pushing machine learning,” the CIO says. “If IBM can figure out how to make a Watson public cloud—that would be interesting.”

When it comes to machine learning, several sources noted Google’s TensorFlow—an open-source software library for numerical computation that makes machine learning accessible to regular users—as a major reason to consider Google. 

And a CEO of a data management provider, who asked not to be named, has partnered with IBM to store its data. The company looked at AWS—it is spinning up sandboxes in the AWS cloud—but for its core business functions, it has turned to IBM Cloud because it didn’t have the contractual restrictions that Amazon was looking to write into the agreement. 

“The thing we found about AWS is that the contract doesn’t lend itself to our kind of customer,” said the CEO. “AWS has language in its contract that they can basically close you down on no notice. If you’re running a kiddie porn site, that makes total sense; but I can’t go to Deutsche Bank and say AWS decided to shut us down. IBM doesn’t have that language. We’re also getting very close with Oracle, though we don’t have a relationship yet.”

This contractual challenge was something that was mentioned by Alex Fuss, director of enterprise architecture for the Office of the CTO at Axa Financial US, noted during that panel at the North American Trading Architecture Summit, that these negotiations can become increasingly complex for businesses that are global.

“This is three years ago, so things might have changed, but when I first joined Axa I was sent to Paris representing the US and there were 55 others representing all different out coves and the first thing that the Europeans told me was, ‘We don’t want to put our data in a US system,’” Fuss recalled, because they didn’t trust that government agencies wouldn’t try to find a backdoor into those European datacenters.

Fuss continued: “Even for more mundane issues, I spent a year working with our corporate lawyers trying to get an AWS contract signed—the initial playing around with the cloud was on some VP’s credit card [to run] a proof-of-concept, but now they were looking to do it at the enterprise level. There were some terms and conditions that they would not agree to—one was that Amazon can shut off a system at a moment’s notice—and I thought it was more theoretical and the lawyers need to dot the I’s and cross the T’s. But I just spoke to someone at a conference two weeks ago and they said that Amazon can do that and they have.”

There Can Be More Than One

According to multiple sources, high-frequency trading (HFT) outfits and prop shops in Chicago have been jumping onto the Google bandwagon thanks to the company’s analytics prowess and strong Chicago-based salesforce. IBM is making inroads with fintech firms, especially those connected to blockchain and distributed-ledger technologies, while pushing its analytics prowess. And Microsoft is leveraging its workflow suite of solutions to bleed into other veins within an organization. Oracle will have to throw money and people at the problem if they want to find any headway into the capital markets space when it comes to its public-cloud offering because it’s lagging far behind, according to those who spoke to Waters for this story. 

As one source from an asset management firm tells Waters, “Why would you want to put all your eggs into one basket? Work with a service provider that best fits your needs. Some do security better, some can support cloudbursting and rapid computing better. The most important thing is to understand what you’re trying to do.”

One might ask: Why make such an effort to gain market share in the financial services sector? The reason—for Oracle and all the others—is that there’s room for multiple winners. A Highlander “There can be only one” situation, this is not. There’s green field to be found. If you’ve built the cloud infrastructure and architected your platforms with cloud in mind, other pieces can fall into place.

It will be incumbent on the cloud providers to continue to show technological improvements to justify further investment—here’s where the arm’s race comes in. Prices will continue to be pushed down, so diversification will come through ancillary products. 

In the capital markets, we’re getting to a tipping point where AWS’ lead is bound to take a hit. But just like Bloomberg in the terminal and communications space, it’s easier to lead from the front. The beneficiaries of this arms race are likely to be the users in the space.