No Integrations: The Future of Cloud
Making two software systems work together is a minor irritant; orchestrating three or more quickly becomes a real headache. The Cloud was supposed to rid us of this meddlesome challenge, reducing client-side interfaces to just one (the browser), and making all the Cloud-based systems communicate smoothly.
Unsurprisingly, this utopian future has yet to emerge. Knowledge workers today operate largely through the browser, and developers are improving user interfaces as rapidly as the technologies can move.
The Cloud – which, let’s be honest, should probably be called the Clouds – simply cannot seem to reach its promised land. However, you combine your Cloud infrastructure, incompatibilities from provider to provider sneak in.
Meanwhile, there is a very active dilemma smoldering at the heart of technology acquisition. Organizations either seek to reduce complexity with a Single Vendor approach; or to maximize system value by opting for Best of Breed.
The Single Vendor approach sounds like a no-brainer: trade marginal best-in-class benefits for reduced administrative overhead.
Except it turns out that most vendors maintain growth after their initial products mature by acquiring other technology. While their business models and systems might merge cleanly together as one, their offerings often don’t, which can adversely impact customers.
The irony! An organization selects a Single Vendor, buys a range of different packages from them – then discovers those systems can’t communicate with one another natively. The compromises, which seemed utterly reasonable in return for a smoothly self-orchestrating environment, now loom large.
The ugliest noun in all of technology is … integration.
An integration is a free-standing piece of technology that helps two or more systems to interact with one another. It can be more or less complex, depending on the level of incompatibility between the connected systems – but it is a source of guaranteed pain over the course of its service. When any of the integrated systems upgrades, the integration requires regression testing. Cloud systems, ironically, increase the likelihood of downtime, because each cloud-based package may upgrade multiple times a year – on their own schedule. With unsynchronized schedules, integrations may need attention six to eight times a year for just a two-system connection.
If the Single Vendor approach requires compromise without a guarantee of full compatibility, are we back to best-of-breed? Given the likelihood of yet more integrations, the answer is, thankfully, no.
Today there is the option of using Platform-as-a-Service (PaaS), a complete development and deployment environment. Architecturally, PaaS sits in between IaaS (cloud-based infrastructure without market-ready applications) and SaaS (discrete software applications hosted in, and accessed through, the cloud).
The prime example of a fully-realized PaaS is Salesforce.com (SFDC). While SFDC sells a full suite of SaaS products, the company also delivers a broadly-available platform upon which other software developers can build and market adjacent solutions. This platform ecosystem unlocks a crucial characteristic: because everything built on it shares a database, core data model, and fully-realized business logic, each individual solution can properly communicate with all the others. In other words, a Professional Services Automation (PSA) application running on SFDC can (within its granted security rights) interact natively with the CRM, and anything else that is SFDC-resident.
When business systems that solve discrete (but connected) problems communicate natively, integrations can be discarded.
The SFDC platform ecosystem has been broadly adopted, meaning Independent Software Vendors (ISVs) have developed a wide range of solutions that can be merged into any organization’s environment. Now an organization can have the best of both worlds: best-of-breed strategy, standardized on a single platform.
PaaS-based solutions fit into a composable technology strategy
The solutions that run on the foundational platform can be mixed and matched over time with lower transition costs than would be experienced with freestanding alternatives.
The alternatives compound already chilling Cloud challenges. The riskiest of all is the Single Vendor-Multi-Platform (SVMP) strategy. A subset of the Single Vendor strategy is rarely selected deliberately but is normally contracted when the Single Vendor acquires new technology that is both (1) in the cloud and (2) based on a wholly different infrastructure. For instance, an organization might commit to a Single Vendor for financials and HR business systems, then acquire their PSA – only to discover the latter uses completely different technologies, and requires integrations, as though it had come from a wholly different source.
In the rapidly-consolidating business systems space, finding oneself unwillingly and unwittingly in the midst of an SVMP environment is increasingly common.
Organizations can protect themselves by knowing the options available through the PaaS solutions already in place. SFDC users, for instance, can explore the AppExchange to discover adjacent business solutions that require no integrations. They should also clarify the infrastructural relationships between their vendors’ offerings: a shared logo does not guarantee smooth, integration-less connectivity.
The future of the cloud is clear: smoothly coexisting systems, seamlessly sharing data and logic, powering continuous organizational transformation. While integrations represent obvious danger zones for bugs and downtime, using the best-of-breed solutions offered by a single PaaS offers the very real opportunity to achieve perhaps the most elusive yet hard-to-attain goal: no integrations.