While it may seem like self-hosting Cube Core is the obvious choice (It’s free after all, right?), it's important to recognize the cost of infrastructure, resources, and maintenance that comes with this “free” choice. Here we attempt to break down those costs to help with your evaluation.
In this analysis we’ll look at three example deployment scenarios for Cube:
- Just Testing (Small)
- Going into Production (Medium)
- Scaling to Enterprise (Large)
But First, Why is Total Cost of Ownership (TCO) Important?
The Total Cost of Ownership (TCO) is a pivotal metric in any technology decision-making process for several compelling reasons. First, it’s crucial to dispel the misconception that self-hosting and internally supporting Open Source Software is entirely free. While self-hosting Cube Core may not require upfront licensing fees, a TCO analysis accounts for all associated costs, including deployment, maintenance, and support, to give a more comprehensive view of what you should expect to spend.
Additionally, a TCO analysis captures long-term savings or expenses by looking at more than just the initial cost of deployment. Oftentimes, streamlined operations can translate into enhanced efficiency and reduced overhead over a longer time horizon.
Finally, most OSS products are lacking important features that can deliver productivity enhancements to the business - further enhancing the benefit of and TCO calculation.
Let’s take a closer look at the three deployment scenarios to understand how Cube Cloud’s TCO stacks up against self-hosting Cube Core.
Scenario 1: Just Testing
When you’re initially exploring and developing with Cube, your query volume is low, your resource requirements are low and you are focused on testing functionality and building a data model that works. Your front-end team will want to test integrations with their tooling and connections from their data apps. It’s also common to utilize a few pre-aggregations at this phase to improve performance for sluggish queries.
On self-hosted Cube Core, you may choose to deploy a single container locally or put it on a small VM so that you can collaborate and your team can connect to it. That would typically take a team several hours or days to learn the in’s and out’s of Cube configuration, then spin up and manage the instance.
On Cube Cloud, you can sign-in to the free tier in seconds which most likely has what you need for an evaluation. Not only does this save you 100% of hosting, setup, and maintenance costs, but you can also get started much faster. When you are ready to go into production, it’s just a click to get a production cluster spun up on the Starter plan. And, even on the Starter plan paying monthly, the TCO would still be less - about half as much as Cube Core.
Note from the Cube team: If you want to evaluate additional functionality from our higher tier plans, just contact sales and we’ll set you up with a limited time trial.
Scenario 2: Going to Production
Launching your first Cube deployment is an exciting and crucial milestone that must be executed correctly to establish trust with your users. You’ll need a stable Cube deployment with high-availability/redundancy for the API and Cube Store worker nodes. You’ll also need a process and tooling for testing and pushing changes to production without taking the deployment offline (CI/CD and blue/green). A Cube Store cluster is necessary to run pre-aggregations and it needs to be sized large enough to execute queries against your largest rollups. The hosting cost is higher than a Cube cluster that only does query pass-through.
On self-hosted Cube Core, you may set up a Kubernetes cluster with 2 deployments - one for Blue and one for Green. Your testing and development work happens on whichever deployment is not serving the production workload. Your CI/CD orchestration pipeline would need to deploy new commits to the non-prod deployment, build or refresh any pre-aggregations in Cube Store that need to be updated, then cutover the traffic routing to the newly updated cluster to complete the code deployment. You likely also need small dev deployments to work collaboratively on feature and dev branches, which requires additional lightweight Cube instances.
On Cube Cloud, you have built-in redundancy and there’s no Kubernetes to setup or manage. CI/CD orchestration runs transparently while you simply merge/commit changes to the main branch. You also have access to chat and email support to help you move faster and more confidently.
Link to production checklist referenced above.
Scenario 3: Production at Enterprise Scale
Things are going well and Semantic Layer adoption is taking off! Cube has become mission-critical to your data teams and products. Data residency requirements also dictate the need for two independent regional deployments. Fortunately, Cube can scale to meet high-traffic demands, insulating your data warehouse from costly overruns and providing consistent data models, access control, cache, and APIs from multiple data sources to many data consumer tools.
With this increased demand comes increased scrutiny. Quick resolution of any data pipeline issues is critical. Infrastructure costs must be both predictable and minimized, but performance and availability must remain high. The stakes are high, and Cube is up to the challenge.
On self-hosted Cube Core, while the baseline functionality scales up well, additional tooling is required to best serve the mission-critical use case. These features would need to be developed and managed alongside a self-hosted Cube Core deployment:
- CI/CD orchestration pipeline
- Application Performance Monitoring stack
- Cluster auto-scaling and auto-recovery
- Responsive, Cube-knowledgeable DevOps support
On Cube Cloud, your deployments are ready to go on day one with all the scaling up functionality built into Cube Cloud already. Clusters can auto-scale and auto-suspend with just a click, and the billing dashboard provides immediate clarity on consumption tracking. Request monitoring is built in so you can see the performance of your API responses as they stream in, and logs can be directed to your sink of choice. Cube Cloud comes with a performant custom-built APM tailored to Cube development needs that's simply not available in OSS. This APM implementation helps Cube Cloud customers dramatically improve their development velocity. You have the full support of Cube’s infrastructure management team and a dedicated Customer Success Manager to keep things running smoothly.
- Self-hosting costs are estimated using EC2 Linux on-demand rates for Scenario 1 and 2 (non-autoscaling). Scenario 3 uses AWS Fargate on-demand pricing for the average cluster size.
- Hourly services rates ($150/hr for small, $200/hr for medium/large) taken from an informal poll of Devops Consultants.
Comparing Apples to Apple Pie
Below is a brief comparison of how we turn the raw ingredient of Cube Core (like apples) into the feature rich, production-ready Cube Cloud (akin to a delectable apple pie).
What If Your Data Must Stay on Your Network?
If your contractual obligation or corporate preference is to retain your data in your own infrastructure, then we have you covered with “Bring Your Own Cloud” (BYOC). Before jumping straight to this option, however, most customers in this scenario choose to list Cube as a subprocessor and can still comply with data residency requirements (e.g. European data can’t leave Europe) with dedicated regional deployments and VPC Peering.
BYOC is simply Cube Cloud deployed in a public cloud account you own and control. The Cube team still manages the infrastructure and supports the deployment, and you get the full benefit of Cube Cloud without any data being stored off premises.
Summary: Regardless of stage or size of deployment, running your Semantic Layer on Cube Cloud costs less than Self-Hosting Cube Core
For dev and test workloads, small production use cases, and enterprise scale deployments, Cube Cloud consistently saves you both time and money. Cube Cloud is supported, fully-managed Cube plus key features to accelerate your development workflow while providing a predictable cost structure. Instead of building out your own self-hosted deployment, Cube Cloud is ready to power your production workload immediately.
We fully believe in the power of open source software and are proud to provide Cube as an open source product. We’ve also created Cube Cloud to provide an even better value to an organization looking to increase speed-to-value, reduce cost overall, and reduce overhead maintaining bespoke integrations. While these are just three common deployment scenarios, there are many other configurations that could be a better fit for your organization. We’d be happy to chat with you about your use case and show you how Cube Cloud is so much more capable and even more affordable than self-hosting Cube Core.