3 Reasons to use multi-tenancy to scale embedded analytics

When product managers and development teams are embedding a cloud analytics platform in their products, there’s a lot to consider. One question that requires more thought than it gets is: “How do we scale when we reach hundreds (or thousands) of users?”

To successfully scale embedded analytics, you need to be able to do the following:

1. First, as the number of customers grows, you must monitor uptime and availability efficiently and reliably.
2. As you scale your apps to hundreds or thousands of users, you must ensure that pushing out upgrades and updates is as easy as when you had only one customer.
3. Your cloud and server resources must be used efficiently across customers, and resources must be pooled and scaled easily.

Therefore, to scale your embedded analytics, the best practice for architecting cloud apps is to design with multi-tenancy at the core.

What is multi-tenancy?

Multi-tenancy in cloud computing is an architecture that securely serves multiple distinct customers with a single instance of a software application. It’s also the best practice for architecting and delivering cloud products at scale. The most iconic cloud companies are all designed from the ground up as multi-tenant. Users run in a single instance while being securely and logically separated. Multi-tenancy differs from single-tenancy models in that the latter requires a separate deployment for each customer.

Example of multi-tenant architecture.

Example of single-tenant app architecture.

Why use multi-tenant cloud computing?

A multi-tenant cloud computing architecture delivers several benefits for vendors―and their customers. For application developers, there are many advantages:

Efficient database upgrades

Managing upgrades and maintenance is simpler in a multi-tenancy model, with just one instance to apply fixes or updates. In contrast, single tenancy requires each customer instance to be upgraded and patched separately, which can quickly become an enormous (and laborious) headache at scale.

Shared computing resources lower the cost

Multi-tenancy enables a provider to allocate shared computing and storage resources across customers, which can result in significant savings. In contrast, single-tenancy requires provisioning compute and storage resources customer by customer. Unlike multi-tenancy, a single-tenant approach makes it hard to align what is provisioned to what individual customers use. Some customers may sit idle—not using the resources allocated to their single dedicated tenant. That inefficiency never happens with a multi-tenant approach.

Observability

A multi-tenant deployment enables a provider’s operations team to easily monitor the health of a single deployment that stretches across all customers and quickly triages uptime issues. In contrast, single tenancy requires monitoring and tracking issues in hundreds of individual customer instances―which is often hard to practically perform without growing a large ops team or significant tooling investments.

Ultimately for providers, it all adds up to delivering software with higher gross margins—there’s less delivery cost while more easily providing stronger CSAT. For example, when customers are all on the same version, that means they have a better support experience. And when operations teams can better monitor uptime, that often translates to better availability.

Putting true multi-tenancy at the center of embedded analytics

So, with all these benefits, if you’re a software vendor looking to embed an analytics platform in your app (which itself is probably multi-tenant), you must ensure that it supports multi-tenancy. This way, your customers can securely run in one deployment.

You may think every analytics vendor’s platform can be run in a multi-tenant model when it’s embedded, but that’s not the reality. Providing embedded analytics to software companies is a side gig for many analytics providers, not their focus. Why? They’re building an analytics platform for companies to use internally, not one designed for software companies to embed and deploy for their end customers.

For many embedded analytics providers, multi-tenancy is incredibly limited. Some offer tenant-only level administration, with a degree of workspace separation, but monitoring the overall environment is painful—creating management issues at scale. While others may offer server-wide administration, it’s challenging to administer at the individual customer/tenant level if needed, making it difficult to customize for specific customers. Often, it’s “cobbled-together” multi-tenancy.

The headaches above motivated the Sisense team to invest several developer decades (yes, tens of developer years) in building next-generation, no-compromise multi-tenancy in Sisense Fusion. If you’re looking to embed analytics at scale, it’s a game-changer. Sisense supports multi-tenancy in several different ways.

With Sisense Fusion multi-tenancy, you can handle all your customers (tenants) from a single, consolidated interface and deployment. You can monitor the health of the entire deployment across all your customers—no need to monitor separate instances. Because all your customers are on one deployment of Sisense Fusion rather than multiple cases, it’s much simpler for your team to perform upgrades and updates.

Monitor all customers from one interface, rather than several.

Next, we’ve got complete logical data and access isolation for each tenant. It means you can run all your customers in a single deployment, separating their data, dashboards, data models, and customizations, all securely and entirely from each other. This makes it simple for you to add or remove customers―no need to provision or de-provision servers or resources.

Choosing a tenant to set model access is simple, with no de-provisioning of servers or resources.

With Sisense Fusion multi-tenancy, you get these management benefits, but we’ve also been careful to ensure no compromise in control. You still get complete granular control, customization, and monitoring of each tenant within the deployment. It adds up to robust tenant-level data, asset partitioning and flexibility, and a multi-tenant model’s efficiency, observability, and cost-effectiveness. It’s the best of both worlds―multi-tenant centralized management, monitoring, and maintenance, with tenant-level security and control.

Multi-tenancy in cloud computing should be at the top of your list when evaluating embedded analytics platforms; its impact on operations as you scale is just too significant to ignore. Making the right decisions about tenancy models is crucial as you embark on your embedded analytics journey, and Sisense is uniquely positioned to help you get there.

If you’re curious about leveling up your embedded analytics, schedule a walkthrough with us.