On-premises data refers to data that resides in a privately owned data center on servers procured or leased by a business. This is the traditional approach to hosting data. Cloud data is an alternative that stores data in a shared data center operated by a third-party service provider.
Why is On-Premises Data Important?
For many organizations, maintaining some of their data on-premises is crucial as they can be confident of its physical location, which many regulations mandate. Many businesses feel they can secure data stored locally better than a cloud. Operational data is often the essence of a company’s market differentiation. For example, Tesla maintains its driving data collected from their customer’s vehicles on-premise because the machine learning models it uses to train are core to their competitive edge.
When data volumes and server capacity are so immense, it can make economic sense not to pay a cloud provider subscription for CPU, network, and storage utilization. Supercomputer farms used by government agencies for high-energy physics, weather forecasting and research are still predominantly on-premise.
Large enterprises such as financial institutions still use legacy mainframe technology in their back-office, and it will take time to re-engineer cloud servers that tend to be Intel-based.
Examples of On-Premises Data
Government Data
Governments maintain data that has to be kept inside national borders. Some cloud providers have created dedicated clouds that can guarantee the security and location of data, but most governments remain on-premise today.
Patient Data
In the Healthcare industry, all personal health information (PHI) regarding a patient’s health and treatment is mandated by the Health Insurance Portability and Accountability (HIPAA). This federal law regulates the protection of sensitive patient health information from being disclosed. For this reason, patient data is predominantly maintained in on-premise data stores.
Financial Systems
Most of the world’s ATM systems still use IBM OS2 for local operations.
Legacy Platforms
Servers based on non-intel CPUs must run on inefficient virtual machines or native hardware that cloud providers do not provide. These include IBM RISC, IBM S.370, DEC VMS, DEC Alpha and HP Itanium.
Real-Time Manufacturing Data
Many of the world’s production systems still depend on AS/400-based JD Edwards software and IBM Series/1 real-time systems that cannot tolerate the latency introduced by cloud-based servers.
IoT and Edge Computing
Edge systems tend to process real-time data streams near where the data is generated. Sensors can include cameras and sensors that need their data filtered and condensed before it is sent upstream. These systems often use mobile computing or embedded processors such as Rasberry Pi, which is not found in the cloud.
Hybrid Cloud
In a hybrid cloud scenario, applications operate on on-premise and cloud-based platforms. The Actian Data Platform can deploy database instances on-premise or in a public cloud. Distributed queries can span both types of database instances transparently to applications.
A newer form of hybrid cloud uses a Virtual Private Network (VPN), which provides the elasticity in server capacity that a cloud provider can deliver and keeps the data on-premise.
The downside of the on-premise approach is that it can run out of capacity. The practice of cloud bursting allows the workload to expand into the cloud when it hits the on-premise server capacity limit.
The Benefits of On-Prem Computing
The growth in IT systems is squarely in the cloud, but on-prem will continue to be commonplace mainly due to the benefits listed below:
- Regulatory compliance is a significant driver for an on-premise system. Organizations know precisely where the data is stored and backed up. Access can be tightly controlled. Systems can be air-gapped, so there is zero external access for really secure installations.
- Cloud migrations can be complex and expensive, so many organizations are happy with their stable on-premise systems. The servers and storage are often paid for and depreciated.
- Existing skills and staff are usually enough to manage an in-house server room for a smaller business, so there is no pressure to rearchitect and learn new skills.
- Autonomy is provided by in-house systems which can operate critical functions locally when networks or HQ systems are down.
- Why rent when you can own? A server that runs on-premise just requires power to operate. There is no need to pay a cloud provider subscription for the work it does.
The Downside of On-Premises Data
Below are some of the reasons why cloud computing is exploding, making on-premise an exception for new deployments:
- Deploying data in the cloud is more cost-effective as the cost per terabyte is low, and there is no up-front capital expense.
- Cloud data is automatically backed up or mirrored to protect it from device failure.
- Server and storage capacity in the cloud is almost infinite.
- Cloud resources are provisioned instantly—no waiting for IT to buy and configure a server.
Actian and On-Premises Data
One of the biggest reasons to entrust your data management to Actian is deployment flexibility, which includes cloud, on-premise, and hybrid options. Try the Actian Data Platform with a free trial here.