SaaS technology is the perfect way to get top applications up and running quickly. If only it were so. Often overlooked is the application development challenge of integrating distinct SaaS applications from an array of providers to create a comprehensive, seamless user experience. The silos associated with traditional on-premises IT haven't disappeared; they've ascended into the cloud.
This story, the first of two parts, examines issues related to SaaS integration. A companion story offers recommendations for conquering those challenges.
Multi-SaaS integration can become necessary for reasons beyond carefully and deliberately chosen SaaS apps from different providers. Line-of-business departments, in their independent good-faith efforts to get their jobs done, may circumvent IT and sign up for the same SaaS app, resulting in multiple instances. Corporate mergers and acquisitions also present a SaaS integration challenge to developers.
SaaS popularity continues to grow
The challenge of multi-SaaS integration is not likely to abate, as SaaS technology adoption continues to soar. In an October 2017 forecast, research firm Gartner predicted that global SaaS-related revenues, which were $48.2 billion in 2016, will skyrocket to $99.7 billion in 2020, a robust compound annual growth rate (CAGR) of 15.6%.
The October forecast is an upward revision of one Gartner issued just eight months earlier. That earlier February 2017 forecast, which predicted a CAGR of 14%, said the adoption rate of SaaS financial applications is expected to accelerate, while the adoption rate of maturing SaaS app categories -- notably human capital management and customer relationship management (CRM) -- may slow slightly.
Allure of the cloud
The initial appeal of cloud computing and SaaS technology -- that they would simplify IT and help cut costs -- has not been proven, at least not yet. "There was the dream, and then there was the reality," said Ed Featherston, principal architect at Cloud Technology Partners, a provider of cloud migration and management services based in Boston.
Efforts to migrate 1990s client-server technology into the cloud revealed, as Featherston put it, "lots of different piece parts out there moving and talking to each other" that were forgotten about. The result, he said, is instead of having an octopus running on IT's existing internal systems, we are seeing essentially the same thing in the cloud.
An apt characterization, that octopus initially consisted of enterprises recognizing the operational simplicity and cost-effectiveness of cloud-based CRM and marketing SaaS technology, according to Carl Lehmann, principal analyst for enterprise architecture, integration and process management at 451 Research.
As a customer analyzes a business's products online, he or she may be offered deals through a marketing tool, engage in a purchase transaction using a credit card and fulfill the transaction. "All of a sudden, we've got a hybrid architecture," Lehmann said, "whereby a business process transcends multiple execution venues."
Graeme Thompson, CIO at Informatica, a provider of SaaS integration technology and tools in Redwood City, Calif., agreed. Following the 1990s client-server revolution and the ensuing wave of early 2000s CRM and ERP provider consolidation -- Oracle acquiring Peoplesoft in 2005 and Siebel in 2006; Microsoft acquiring Great Plains Software in 2001 -- the allure of the cloud became irresistible, he said. "For the first time, we could consume these things simply," Thompson said. "They were easy to deploy. There was the lure of quick time to value."
The inevitable consequence, Lehmann said, is the success of these disparate cloud efforts resulted in systems that required linkage. Once businesses started maturing their workloads in public clouds, they found themselves consuming more and more public cloud services, he said. Conduits to glue the pieces together were the solution for IT.
Ultimately, a user issue
For users, the issue is not one of building conduits, but rather seeing the aggregate result presented in a simple, comprehensible way.
Randy Heffneranalyst at Forrester Research
Juggling multiple SaaS apps to accomplish a single business task can be head-spinning, if not implemented through proper interface design. "A customer service associate may have to look up an order in an eCommerce app and switch to a marketing app to verify details of a promotion, then to a financial app to issue a customer credit, and finally to a customer relationship management app to record the details of the interaction," Randy Heffner, a Forrester principal analyst serving application development and delivery professionals, wrote in the April 2017 report "How To Set The Right Strategy For SaaS Integration."
"No application vendor is going to give you everything that you need in order to run your business front to back, top to bottom," Heffner said in a separate interview. The allure is that SaaS makes the promise there's nothing to install, an idea that harkens back to Salesforce's original premise of no software, he said. "That sounds great, sounds easy; let's just get something and go."
It's not easy, and software decision-makers are concerned, with 52% noting that SaaS technology application integration is a challenge, Heffner said. That concern encompasses not just SaaS-to-SaaS integration, but SaaS-to-legacy homegrown apps, as well.
Even at the highest levels of Oracle, a major cloud services and platform provider, the concern about the challenge of SaaS application integration is clear.
Ashish Mohindroo, vice president of Oracle Cloud, said beyond the proliferation of cloud services common in businesses is those services are often built atop different technology platforms and do not use common APIs that would ease integration. IT's history of purchasing siloed applications has transformed to a similar scenario of businesses still consuming siloed services, but in the cloud.
And while we'd all like to think developers solve the integration challenge simply through integration platforms and APIs, that is an oversimplification that ignores real-world challenges related to performance, network latency and data residency, according to Michael Day, principal database engineer for cloud services at PTC. Based in Needham, Mass., PTC itself has morphed from providing product lifecycle management services to melding internet of things and augmented reality to help companies transform their own product design and operations processes for the digital world.
"We have processes and database services that are compute-intense, with lots of I/O," Day said. "If that is on site, you have no latency issue. But when you migrate applications to the cloud, you need to account for that."
Milliseconds, perhaps, but when large amounts of data are being moved, in a weather analysis system, for example, those milliseconds between applications running in different clouds in different geographies and on different technology stacks quickly add up, Day said.
In lockstep with database performance is data governance. "If you're going to move data around just to make sure that it's secured, "Day said, "you have to make sure that you catalog it so that you know where the data exists, where the data was created, who has access to it and where else it's used." Think of it as cloud silo management.
Swipe and go
A key appeal of SaaS technology is signing up takes just minutes and requires nothing more than a credit card. "You wind up often having multiple buyers within the same organization who jump on this SaaS bandwagon to get something for their departments," Heffner said.
According to Heffner, other Gartner analysts said it's not unusual to see enterprises running as many as 15 distinct instances of Salesforce simultaneously. Multiple instances of any one SaaS application would, by definition, be running on a single technology platform and use a single set of APIs. The inevitable discovery of such a situation -- perhaps by a CFO responsible for paying the monthly bills -- may ultimately require consolidation by the organization's application development staff.
The problem of multiple instances that eventually require integration is not new. In 2016, Liz Herbert, a vice president and principal analyst at Forrester Research, noted that businesses expect to undertake integration projects for CRM, fulfillment, payroll, inventory and other unrelated systems. It's those dozen instances of Salesforce, which were each brought in "under the radar," that are often the result of line-of-business departments acting alone, she said.
Mergers and acquisitions constitute yet another scenario in which multiple SaaS applications require integration. Even if the acquiring and acquired organizations use the same cloud-based SaaS payroll service, it's a good bet that the implementations are not identical. It will be up to the acquiring company's application developers and operations personnel to analyze the database structures of each and remap fields to bring the two into sync.
In addition to Informatica and Oracle, other tool providers are scrambling to help businesses bring the pieces together. Dell, IBM, MuleSoft, SAS and TIBCO are among the larger players, with Cloud Elements, Jitterbit, Kloud Connectors, OneSaas, Scribe, SnapLogic, Tray.io, Zapier and others part of an increasingly crowded field.
Low-code tools can ease SaaS integration woes
Build SaaS integrations using up-to-date developer skills
SaaS integration is a never-ending priority