Photo: The engineering team behind Elastic Hypercube Technology.
Today, we’re proud to launch our patent-pending Elastic Hypercube Technology. For customers, this breakthrough represents an entirely new level-set in their business planning. For the team that made it possible, it was a labor of love, passion, and many sleepless nights.
There’s a story behind why we worked so hard for so long on bringing Elastic Hypercube Technology to market. If you have a few minutes, we’d like to share it with you here.
In 2017, our customers helped us articulate why the old ways of doing planning, reporting, and analytics—manual, siloed, and error-prone—were static and therefore not having a meaningful business impact. Conversely, an active planning process—where planning is continuous, collaborative, and comprehensive—could help businesses make smarter decisions faster and truly drive business agility and value. This wasn’t exactly breaking news to us. Our founder, Rob Hull, had been evangelizing the roots of this concept since the company’s very first days.
So, in 2017, we made a promise to our current and future customers that we will continue to innovate and to deliver a no-compromise platform. One that is easy so everyone in the organization can collaborate, powerful so you can model anything and analyze everything, and fast—fast to implement and fast to evolve.
When it comes to enterprise software, it’s easy to make promises. Keeping them is a lot harder. So we had our work cut out for us.
This was especially true because planning systems have always had challenges dealing with large and/or complex models. These challenges have led software providers to force their customers to accept compromises on the user experience. Users have had to live with an array of limitations on model scale and flexibility, slower analytics, and offline storage of older, less used calculations. Providers have also put ceilings on the number of scenarios, constrained dimensionality or size of dimensions, required additional cubes for just reporting or analytics, imposed restrictions on how detailed the calculations can be, and more. Of course, these limitations don’t impact all users, but they happen more than they should, even for state-of-the-art, modern cloud planning platforms like ours with thousands of customers. And looking forward, we see how business data is nearly doubling every year, and how organizations want to involve more business users and functions in their planning.
Considering these trends, it’s natural to wonder: What might these challenges look like next year? We weren’t about to wait around to find out. We had made a promise, and we were going to keep it.
Introducing Elastic Hypercube Technology
And so was born Elastic Hypercube Technology, the industry’s first technology to address these challenges head-on. It’s a next-generation planning, reporting, and analytics engine that automatically scales with the demands of a modern business.
Elastic Hypercube Technology imposes no limits on the number of dimensions or dimension values. It allows you to model as many scenario versions as you can possibly conjure up. It erases guardrails on data set sizes or model complexity. It enables you to report and analyze without premeditation. It intelligently keeps at your fingertips the data you need all in real time, and it scales to whatever heights you need. It’s the engine that drives our Business Planning Cloud platform.
And it does all of this without sacrificing the ease of use and flexibility Adaptive Insights is famous for. In fact, to users, this all takes place under the hood. The only things they’ll notice are faster calculations and even more flexible, no-limits modeling.
To achieve all this, we rebuilt our platform in four key ways.
We made the whole architecture scalable
We took what had been a monolithic application and split it into core services, starting with a compute service and a metadata service. The compute engine is now a separate yet tightly integrated component of the platform. Because our users compute and view a lot of data, we ensured these services could communicate with each other through highly optimized protocols.
With a horizontally scalable architecture, we can now transcend the traditional CPU and memory limits of a single server by distributing workloads dynamically across server farms. This allows us to give models the compute and memory resources they need when they need them. We deployed this infrastructure on Kubernetes with dynamically sized Docker containers on both bare metal and Amazon Web Services.
All this buys an immense amount of scalability. How much? Do you want to do personnel planning for 100,000 employees? Go ahead. Want to run a report across 10s of scenarios on an M&A you are evaluating? Have at it. Now the entire architecture scales to the size and complexity of your problem.
We made our calculation approach super intelligent with optimized calculations
A lot of time and resources tend to be wasted when a user makes a small change to a model, because traditionally, the platform would recalculate much more than is needed in the model. As time-consuming as these calculations can be, the geeks among you already know that a dependency graph of a sparse, complex, multidimensional space takes more time to manage than to actually calculate.
We knew we had to fix this. So we applied our advanced understanding of model dependencies to invent the industry’s first fine-grained dependency tracking algorithm and data structure. With it, we can do data tracking much more efficiently. Now when something in a model changes, Elastic Hypercube Technology recalculates only the changed elements of the model. Now a large restaurant operator can efficiently analyze the impact to profitability of rising labor costs to total labor expenses, COGS, gross profits, and more. By focusing only on calculations that are truly necessary, the platform enables faster what-if scenarios.
We turbo-charged it with intelligent dynamic caching
We knew we couldn’t sacrifice user experience (response time, ad-hoc analytics) or modeling flexibility even as we delivered next-level scalability, because for SaaS applications, response time is everything. So we made it possible to introspect the model and its behavior at run time, then apply cost-modeling and optimization techniques. The result is intelligent dynamic caching. The engine improves and optimizes its caching technique by observing and responding to run-time activity.
In other words, the intelligent engine learns and optimizes on the fly what it needs to cache, what it needs to calculate, and when in parallel. Now a retailer with a million SKUs can forecast sales of each SKU by dimensions like location or sales channel. And it all happens under the hood, transparent to the modeler or her business partner.
We built upon the flexibility we have been always known for
We let you model what you want and analyze as you want. You continue to have the unique ability to model (and view) multidimensional data in tabular or cube format. As an example, a company can model personnel or contracts in a tabular format while exploiting its dimensionality and model revenue as a cube with dimensionality—then link all these components together in an overall model. The Hypercube is available for ad-hoc analysis and reporting, requiring no cumbersome setup or additional cubes.
A new level-set for planning
It’s not easy to solve problems that have never been solved before. But with a great team (more on that in a future blog), anything is possible. Our labor of love, passion, and many sleepless nights is launching out in the wild. It’s the Elastic Hypercube—intelligent, scalable, and flexible; all without compromising ease of use and time to value.
To be able to continually innovate, without compromising the things that help businesses of all sizes to make smarter decisions faster, is a big promise to keep. Today, a new generation of planning platform is born, and with it we’ve made another big down payment on our promise to our community.
It is easy to make a promise, but it’s really hard to keep it. We’re proud of how we made good on this one. And today, we’ll make another: We plan to continue to innovate without compromise, because that’s what our users need.