Author: Steve Lock
The key differences between on-prem and cloud-based data platforms are the hardware and provisioning of systems.
This article details the key differences between these platforms, and discusses some pros and cons to consider before you select the right approach for your organization.
On-Prem Based Systems
On-prem solutions run on your own architecture, and require purchasing servers and resourcing with the appropriate technical staff. It’s important to keep in mind that managing and keeping the systems running sits mostly with you instead of a vendor or contractor partnership.
This option also includes significant upfront costs, such as purchasing the hardware, and depending on the chosen system, you may need to purchase more traditional software licenses with higher initial costs as compared to SaaS.
On-Prem Advantages
On-prem can be seen as an antiquated approach, but my hunch is that we’ll be seeing more organizations move back to on-prem. Expect it to become more trendy. Key advantages include:
- Some organizations report much lower costs. An interesting example is Ahrefs, who claimed to save $400M over three years.
- You gain full control of your servers, whereas cloud vendors often reduce control, especially with managed services.
- Some businesses like to own all their data and systems, reducing third-party/vendor access. Healthcare is a common industry example of this.
On-Prem Disadvantages
Some key disadvantages include:
- The biggest is probably the upfront costs. Be prepared to pay far more and wait a number of years to recoup advantages. This alone makes this option prohibitive for smaller organizations.
- Specifically for data, on-prem management is often seen as inferior to cloud-based solutions. Modern data analytics often pushes for cloud-based systems as the primary approach.
- Historically, on-prem has been perceived as requiring more in-house staff, although some organizations report the overhead is quite similar for cloud computing. Either way, maintenance and dealing with hardware failure can be much more painful with on-prem solutions.
- Clients are on the hook for implementation, so your mileage may vary depending on the quality of your technical teams.
Cloud-Based Systems
Cloud-based solutions are where architecture and computing resources are ‘rented,’ typically from major cloud vendors. These solutions often include managed services tasked with keeping all systems up and running.
Currently, there is much discourse around cloud-based systems having hidden costs, including requirements to hire additional engineers that might not have been budgeted for, and estimates around the pricing coming in higher than expected.
Cloud Advantages
We’re expecting to see more high-profile examples of major organizations moving back to on-prem, although there are also some major advantages of cloud-based systems including:
- Lower upfront costs.
- Can be ideal for startups and smaller orgs because you can start fast and with lower budgets.
- Cloud-based systems are typically seen as far superior for advanced data analytics, and migrating is often put into project plans as a prerequisite.
- Major benefits for disaster recovery (DR) because when hardware fails, it’s handled by the cloud vendor.
Cloud Disadvantages
Some key disadvantages include:
- Vendor lock-in. It can be particularly expensive and painful to migrate a larger organization if all of your data sits in the cloud. (It’s common to experience low costs to ingest data and much higher costs to migrate the data.)
- More limited control. I worked on one project where a database couldn’t be fully accessed on managed service for critical maintenance, but in an on-prem scenario, they would have had full control.
- Reports of higher than expected bills versus business planning.
- Some organizations call out that more engineering resourcing is needed than expected.
- Complicated billing and cost estimates. Most organizations have stories about exceeding limits and receiving higher than expected charges, such as using too many IP addresses, large queries, or increasing the number of secrets stored. The TLDR is that there are often limits that can be easy to exceed and working with billing data can be much harder than expected.
In conclusion, cloud-based solutions have been perceived as superior and cheaper in recent years and have been by far the trendier option, too.
However, over the last few years, the gap is closing and we’d recommend not rushing into either option before assessing the value of both possibilities for your specific analytics context.
PS. If you’re considering hiring a Data Analytics expert to help determine whether you’d benefit most from an on-prem vs. Cloud-based analytics platform, don’t sign a contract until you check this out: 4 things to know before hiring a Data Analytics expert.
