When On-Prem Cannot Replace IaaS

When cloud costs from hyperscalers spin out of control, a common optimization question raised in the board room is whether to move these cloud workloads to on-premises. While you can certainly cut costs this way, such a move will greatly impact your tech stack and must be planned very carefully. 

Cloud-based infrastructure isn’t just a financial exercise that switches CAPEX for OPEX. There are some inherent advantages to using an IaaS provider that cannot be replicated by running infrastructure on-premises. Typically, these include:

• Latency sensitivity
• Global expansion and delivery
• Multi-user scalability and concurrency

As such, if companies require any of the above, they will most likely have to opt for an IaaS solution. Let’s explore each of the above:

Latency Sensitivity

This refers to use cases where individual users have little tolerance for latency, such as real-time applications, which can include video processing at the edge, autonomous driving, virtual and augmented reality, remote control of IoT devices, trading and retail, and the like. 

It includes:

• Network-induced Latency - consists of the roundtrip latency from a user making a request to the place where it is processed.
• Compute-induced latency - represents the delay associated with processing a request at an infrastructure level, such as cold starts. 
• Storage-induced latency - refers to the processing time associated with retrieving data from or writing data to a storage system, such as using KV stores, in-memory databases, and others.

Latency sensitivity also encompasses multi-user scalability and concurrency, which includes instances where the volume of request and time spent processing each of them will have an impact on the overall experience and the scalability of the solution.

Global Expansion and Delivery

There are instances where an organization is rapidly entering new geographies and requires a global infrastructure provider to support their expansion.

• Geographical presence - typically represented by the number of data centers or points of presence a provider operates in different geographies. These need to be close to major metropolitan areas across continents.
• Legal, compliance and regulations - refers to the legal considerations a customer must work through to operate in that geography. This can include setting up local entities such as limited companies, working with local providers, and even data sovereignty requirements.
• Programmability - is about interacting with the underlying IaaS solution for provisioning and configuring the infrastructure at scale and dynamically without tedious GUI menus..
• Caching and storage - refers to the ability to distribute and store content closer to end users, reducing retrieval times and origin server load. Effective caching strategies dramatically improve performance while reducing bandwidth costs and origin server load.

Failover and resilience - this ensures continuous service availability even during component failures, traffic spikes, or regional outages. It includes redundancy mechanisms, high availability configurations, active-active configuration, Anycast-based networks, automated failover processes, and disaster recovery capabilities across distributed infrastructure. 

The Sweet Spot is Between On-prem and Hyperscalers

The good news is that your organization is not in an either/or situation, where you either go full-cloud or full on-premises.

You can have the programmability of cloud and the lower-level control of on-premises deployments by using NetActuate’s Open Network Edge ONE).

NetActuate is a global provider of managed network and infrastructure services, offering a third option to the otherwise binary cloud/on-prem question. Open network Edge is a globally distributed infrastructure-as-a-service (IaaS) that provides out-of-the-box implementation of widely deployed open source software, including operating system, network operating systems, monitoring software, orchestration tools, and everything else needed to scale your platform.

To get a framework for evaluating global infrastructure options, and learn more about the Open Network Edge, download our latest ebook, Exit Hyperscalers. Keep the Cloud.