In cloud computing, people often use the terms IaaS (Infrastructure as a Service) and PaaS (Platform as a Service) interchangeably, but in reality they are completely different ways of distributing virtual services.
At a high level, this might be equated to renting an empty home (IaaS) or furnished (PaaS) – and how one utilizes the features depends on the individual. Most homes will have a bed, table and chairs – but who purchases the components, maintains them and how the homes are utilized may be vastly different.
IaaS is a primary services model for public cloud providers, and offers users a virtual machine undefined for any specific type of software, raw block storage for applications, load balancers or networks. Cloud users utilize virtual machines, installing operating systems or application software and then the end user is responsible for the configuration and management of the software and is billed on utility computing approach (resources consumed).
Under the PaaS model, cloud providers deliver a computing platform or solution stack (MSQL, Tomcat, etc.) that will have an operating system, programming language execution environment, database and/or web server included. App developers can run their software on top of these platforms without building and maintaining them directly. PaaS environments are also priced by usage, and can also scale automatically to match increased demand without changes to the platform.
Service Approaches and Cost Implications
We have studied each service approach and understand there are similarities and differences in managing the cost impact in IaaS and PaaS environments.
In terms of similarities, Cloudyn’s recommendation engine is able to reduce over-provisioning or under-utilization of resources in for both PaaS and IasS models with a combination of rightsizing and price optimization. While PaaS models typically have fewer pricing levels – there is still the ability to measure utilization of the services and make clear recommendations of how to reach a balance between performance optimization and cost savings.
But the two service models also have some stark differences:
- Performance Metrics from cloud providers:
- IaaS performance is largely available in raw data form from the cloud provider and is measured primarily by a few simple parameters: time and memory/resource utilization. Direct access to this data can enable the user to analyze and structure the overall parameters that are directly relevant for their business, but there is little guidance in the analysis of this information from a cost perspective. For instance, Amazon Web Services allows customers to gather CPU utilization, Network and I/O.
- PaaS reporting can have as many as 20 parameters to maintain production environments that offer the performance requirements necessary for the application (average utilization, max memory, summary of input and output measurements, etc.) and the right threshold is necessary for the service capabilities to remain intact. On PaaS the user has no access to the box and that’s why he can’t extract additional metrics by himself. The cloud provider is collecting and delivering the data – and then it is up to the user to determine which information is useful in measuring costs and provisioning models. For instance, Amazon Web Services RDS allows customers to gather CPU utilization, Memory, Latency and throughput.
- IaaS has flexibility to become whatever type of service necessary for a business, and the pricing models reflect the possibility for multiple approaches.
- For Paas, particularly with Amazon RDS, the models are more rigid in configuration and the variable costs.
Why does this matter?
Cloud economics drive the cost impacts on business environments – and the ability to identify, analyze and configure optimal cloud resources continues to increase in complexity.
Each business has reasons for selecting the particular model (IaaS vs. PaaS) based on internal expertise, technology requirements, software preferences or any other number of factors, but it is important to realize that cost optimization can also be an aspect to consider when in choosing one approach over another. Matching your business objectives with the right services configuration can also save time and money in the management of your cloud deployments.
No matter your choice, with the visibility and recommendations available from Cloudyn, businesses can realize real savings from the cloud in both IaaS or PaaS environments– illusive savings for many businesses today.