In a linkedin post by Fergal Mcgovern in May, he tries to explain why around 83% of enterprise CIOs plan to place some workloads on-premise instead of on-cloud. Let’s briefly explain what we mean when we say on-cloud and on-premise:
On-cloud storage is when data is stored in data centers operated by third parties which sell storage and compute options. The three main notable ones are Azure, Google and AWS. You basically store models and data on remote servers which are controllable and accessible via the internet.
On-premise storage is when data is stored in a data center owned or operated directly by your company. As opposed to on-cloud storage and compute, you do not necessarily need internet access to operate.
There are many reasons why this could be happening. In this article, we will expand on Ferhal McGovern’s answer which is linked to the increased use of AI. We agree with this assessment.
The increased use of AI
AI use is becoming more and more widespread recently and is becoming more rampant in many industries. According to the State of AI released by Mckenzie in May 2024, details how much AI adoption has surged. From being considered a more experimental technology to actually being used for concrete production use.
With this increased usage, increased complications are seen as well. We will delve into the tree main ones: Privacy and data protection, increased regulatory constraints and cost saving measures.
Privacy and data protection
As more and more data is stored in the cloud, data protection from breaches and malicious third-party use gets more complicated. In a guide released by Cloudian, they detail some of the data protection issues faced when deployed data and models on the cloud.
Firstly, third-party hosting limits visibility into data access and sharing. When companies use cloud companies to host data, they lose the ability to directly monitor who can see, share and edit that data. They have to rely on trusting the cloud provider. According to ProcessUnity, the average organization uses 6000 SaaS service third parties to store and manage data. The sheer amount of data and trust you need to have makes this risky.
Along with having to protect your data from third party malicious usage, the risk of breaches and leaks also increases with public cloud usage. The amount of data breaches over the years is countless. Arcserve listed the 7 most infamous breaches which have occurred.
In general, using public cloud means trusting third party providers with storing data. In data sensitive industries, this can increase risk and lead to complications down the road.
Increase regulatory constraints
Along with risky data storage problems, regulations have also become more stringent recently. The GDPR rules have been in place in the EU since 2016 and have stringent requirements for data storage and use. Being GDPR compliant also means that the entire data pipeline has to be GDPR compliant. If the cloud provider’s servers are in the US, this raises questions about compliance.
Along with having to follow GDPR, AI users will now also have to be compliant with the EU’s AI act. We wrote an article surrounding data protection and privacy in the healthcare industry in the Netherlands. In general, having to comply with these regulations along with having data stored in the public cloud will be a headache which can be easily solved by storing data on-premise.
At UbiOps, you can use our platform on-premise, but if you opt for cloud, you can be assured that we store all our data in the EU region.
Thomson Reuters released an article about the risks of storing data on a public cloud. Cloud security is a shared responsibility, requiring legal teams to evaluate the cloud provider’s security policies, ensure communication throughout the data oversight process, and maintain audit trails to detect unauthorized access.
Costs
Finally, storing data on-premise could also have cost saving benefits. While storing data on-cloud can save costs initially, if AI and data storage needs increase, the costs, along with license and usage fees, will also increase. In the long term, it can be less costly to store data on-premise. Helixstorm details this in their blog, details how Internet usage costs go down when storing on-premise.
Conclusion
In conclusion, companies are increasingly switching to on-premises storage instead of on-cloud. This is for 3 reasons. Firstly the increased use of AI in data sensitive industries. Secondly the risk of data leaks and breaches and finally the long-term cost-saving effects it can have.
UbiOps is well positioned as an AI platform to be used as an on-premise toolkit. We offer the ability to deploy on-premise with full functionality. If you are interested in our platform, contact us!
