We are excited to bring Transform 2022 back in-person July 19 and virtually July 20 – 28. Join AI and data leaders for insightful talks and exciting networking opportunities. Register today!
Is there a way for businesses to bring order to the burgeoning world of microservices, APIs and databases? Today, Apollo is tackling that problem by launching Supergraph, a software layer designed to unify all of the disparate data feeds and create one source for front-end developers.
Supergraph is designed to offer a one-stop source for data and data processing. It is a composable collection of tools that can work together to generate reports, dashboards and general answers. The constellation of tools builds upon the increasingly popular GraphQL protocol and opens up access to a large-scale enterprise while ensuring security and privacy.
“We call it the Supergraph because it’s a graph of graphs.” explained Geoff Schmidt, the CEO and cofounder of Apollo. “We think of it not as a point solution to connect the front and a back end, but as a composition layer. It’s a new layer in the stack that sits in your enterprise.”
Keeping businesses up to par
The new platform builds upon Federation 2, a GraphQL-based system that was launched in April. The system collects data across the different data sources with a simpler syntax designed to save developers and average users from the complexity of data types, directives and keywords.
“This lets businesses keep pace with the rate of change,” said Schmidt. “When something like COVID-19 happens, you need to be able to quickly reconfigure your business to keep pace with new opportunities, changing markets and changing needs. “
Schmidt estimates that some of their best clients have been testing the product and changing the structure of the Supergraph as often as 30 times a day. New APIs and data sources are added and subtracted quickly as developers add new features and incorporate new data flows.
In essence, Apollo wants to bring some of the flexibility of agile development to the APIs themselves. It wants to engineer them so they’re flexible enough to change frequently and keep running. Developers will be able to fold in new features without jeopardizing running code.
“The ways that don’t work are if you bring in an academic ontologist, or you bring in a management consultant and say, `Hey, I want you to interview everybody at my company and build a UML model of everything inside my business in some modeling tool,’” said Schmidt. “That’s gonna be out of date the minute you finish. That waterfall approach doesn’t work.”
Increasing challenges for developers
Making sense of the increasing complexity of microservices and APIs has been a growing challenge for developers. Tools like Swagger from Smartbear Software and Kong are just two examples of popular API management tools that both document and add structure to the various microservices that are becoming common.
Many clouds are also offering API gateways that can offer much of the same functionality with additional help regulating access and providing security. AWS’s API Gateway, IBM’s API Connect and Google’s API Gateway are just some of the options for regulating the flow of data in and out of the microservices lying behind them.
These API gateways speak a variety of different protocols. Apollo wants to capitalize on the growing interest in GraphQL and its ability to provide a concise and easy-to-understand query language that is still powerful enough to specify a very diverse range of data. A number of databases like Fauna, MongoDB or Yugabyte already support GraphQL directly. Others like PostGraphile, Prisma and HyperGraphQL are creating GraphQL interpreters that work with traditional databases.
“Many companies are drowning in a sea of complexity and looking to streamline a modern-day technology stack,” said Mike Leone, senior analyst at Enterprise Strategy Group. “The Apollo Supergraph does just that in a novel way. The platform empowers teams to quickly integrate modular data, services and business logic into a unified, relevant and timely customer experience. With the Supergraph, Apollo has also anticipated and addressed issues relating to federation, security, validation and scalability as organizations’ graphs grow over time.”
Apollo wants to build a more sophisticated layer that can act like a meta API layer that integrates the results from any number of services and databases that live behind it. This simplifies life for the front-end developers who will have one-stop access to any of the data. If the information evolves, they’ll still turn to the same Supergraph but add new parameters to ask for the options.
One of the challenges is adding layers of governance to control access. Apollo believes that their tools offer a declarative model so developers and stakeholders can set clear rules that control access to certain sections of the graph.
“The advantage of this declarative architecture is you see every query going through the system and you can apply a consistent set of rules and completely consistent set of monitoring or analytics to it.” explained Schmidt. “So you can know exactly which piece of data comes from where and where it went to and who authorized that and why you can set those rules on a forward-looking basis.”
The Supergraph bundles together several Apollo products like Rust Router and Studio. Several new features and enhancements to them are also rolling out today. Rust Router, a GraphQL query processor, will now be open sourced and available for download. Apollo anticipates that it will deliver substantial improvements over previous options.
Studio, the development environment for building GraphQL queries, will now be able to provide more build-time bug checking by comparing queries against schemas. The basic version will now gain more advanced features that provided feedback on schema checking that were once only available to enterprise customers.
Apollo plans to roll out much of the standards and some of the background software as open source. The Studio, though, will be available with a free tier and a paid tier that supports the enterprises building elaborate Supergraphs.
“We’re building a full declarative query planner and then really mapping out what all the change management and metadata the developer needs,” concluded Schmidt. “The great thing is this is now proved on a large scale.”