Cloud is no silver bullet
The RISE Private Cloud programme demonstrates that monolithic SAP systems aren’t going extinct any time soon. For a specific group of customers – both new and established – SaaS will never be an option, thanks to the fine control and flexibility an on-premises system offers.
ECC and S/4HANA on-premises ERP systems require detailed control over the hardware and operating systems they run on. Because of this fact, some of the greatest and cheapest features offered by cloud service providers (CSPs) – think serverless computing, pod-based processing – are null and void. There’s no real difference between maintaining your own hardware or paying a CSP to host it; IaaS is, in essence, just somebody else’s equipment.
When it comes to security and stability, a simple move to the cloud isn’t going to offer earth-shattering benefits. It’s no hassle to design a stable architecture in the cloud, but connectivity between its integrated services must also be guaranteed.
Thus, a current SAP infrastructure model in the cloud does not necessarily offer a conclusive solution for everyone's desired level of control and flexibility. But are there other aspects to consider where on-premises hardware could offer an edge?
Flexibility-related reasons to stay on premise
Opex vs. capex
The opex IT budget model offers clear benefits, but it does mean that an organisation no longer has the option of postponing IT costs if the budget is tight in the short term. The risk is having your infrastructure frozen by your CSP. Even with the pay-as-you-go model of CSPs, keeping specific virtual machines (VMs) powered down doesn’t mean you get to avoid paying for their associated OS licenses and storage costs.
Buying your own hardware means that it’s yours to use as you see fit – even if that means going for extended or third-party support.
CSPs do, if rarely, experience resource shortages, making it impossible to spin up new resources of affected types. This can last minutes, days or even weeks. Although production workloads aren’t typically hosted on VMs that are restarted regularly, non-production systems can suffer as a result – a risk avoided if the hardware is firmly under your control.
Relying on on-premises hardware means that there are fewer links in the critical chain of connectivity, and, importantly, fewer links that aren’t under an organisation’s direct control. A stable internal network with enough bandwidth can be enough to keep core processes on critical systems running.
Monitoring and performance troubleshooting
To reap the benefits of IaaS-level economies of scale, there must be a vast structure in place for an organisation’s own components to plug into – and these are rarely right next to each other.
Having your own infra setup makes it easy to pinpoint which physical machine is running which application, which network kit covers communication, geographical distance between components – and component status. You’d have to jump through quite a few hoops to get that information from a CSP.
CSP’s offer quite a lot of flexibility in terms of provisioning new hardware, no doubt. But where hyperconverged infrastructure and local hyperscalers beat their SaaS cousins is in advanced functionality and resulting flexibility. Hyperconverged infrastructure offers a platform in the long term with spare memory bays, storage slots and additional CPU sockets. These specific components can be obtained quite quickly to tackle resource needs or dynamically added with a hypervisor.
Cost-related reasons to stay on premise
Estimates and sizing
Are you fully aware of your systems’ dependencies? Do you know how much processing power is required by your applications? How much data you have stored? The amount of bandwidth your data centre uses monthly? How these values are likely to evolve over the next few years?
Moving to the cloud means paying for not just processing and memory for the VMs you build, but for storage, bandwidth, etc. If your organisation is not familiar with the technical aspects of its IT environment, unexpected costs might arrive on your bill from the CSP.
Although it’s tough to estimate the size required for an organisation’s physical machines (particularly over a 3-year period), the same applies to VMs running in a hyperscaler. Constantly running VMs are the opposite of cost savers in the pay-as-you-go model.
Within that price model, you could rely on reserved instances (RIs) to grant you additional scaling opportunities. However, as RIs are tied to very specific VM sizes, a proper estimate upfront of your hardware requirements remains essential to meet scaling and budget targets.
IT department needs
Every organisation will need IT personnel or an IT department. Outsourcing your IT activities to a third party only changes the origin of the invoice.
In moving to the cloud, the first thing your organisation will need is an individual who is skilled in the setup and migration of your landscape – and once there, you’ll need trained personnel to operate and maintain your cloud environment.
When it’s easy to spread out, there’s a risk of investing in unneeded resources. This is a big enough problem that CSPs cut into their own profits to offer free tools that help organisations counteract it and take active steps to limit it.
Businesses that stay on premise don’t pay anything extra for new sandboxes or VMs that run on their owned hardware. Even more, this model offers an additional buffer of approvals/awareness, preventing those deemed unnecessary from ever being created.
Prepare for all contingencies – in the best interests of your organisation
Together, hyperconverged infrastructure and hypervisors offer a key solution to technical hosting requirements, particularly when an organisation needs advanced technical functionalities and/or firm control over its systems.
Convinced that that running S/4HANA or ECC with on-premises hardware is right for your organisation but want to keep your options open?
An established footprint with a CPS of your choice – for some solutions requiring full network integration – could be a viable choice, empowering you to hit the ground running if a temporary issue comes up.