IaaS vs PaaS vs SaaS:
A tale of two Zoos – Great Legacy and Futurama

Introduction

Software has eaten the world; more and more of what companies do, depends on software. Software manifests itself to users as applications and many of these applications run on servers in data centres. One would expect that the number of servers a company is running scales proportionally to the number of applications used. After all, servers are the horses pulling the application carriages.

Sadly, there is hardly a relationship between the applications used in the company and the number of servers it operates in the data centre. Over time, the number of infrastructural and maintenance functions required in a data centre has exploded and most of these functions also need to run on servers. Consider the server footprint of a medium-sized organisation: 65 out of a total of 152 servers were of an infrastructural nature. Such statistics are in no way atypical.

Application servers are a bit like an exotic animal in a zoo. In order for the animal to live and prosper, a large number of supporting functions need to be put in place: living quarters, dietitians, vets, breeding specialists and food suppliers etc.

A tale of two zoos – Great Legacy and Futurama

Once upon a time, there was a city with two great zoos: Great Legacy and Futurama.

Great Legacy had a long illustrious history and believed in organising everything in-house. As the CEO of the zoo bought more exotic animals to increase the attractiveness of the zoo, the operational strain grew. Each animal species required specialists: a vet, a food supplier, appropriate living quarters. The introduction of a new species implied the recruitment of the right specialists or the training of existing specialists in the particulars of the new species. Finding and training specialists for exotic animals was time-consuming and expensive. Unfortunately the CEO expected all this to happen quickly and efficiently – Futurama, the competitor, kept broadening its assortment of animals at breakneck pace.

Futurama was a younger organisation and it had realised that it was too small to organise every aspect of its business on its own with high quality and efficiency; this was especially the case for exotic animals. For this reason, it had contracted Zoo Platform Services (ZPS) as a supplier. ZPS serviced thousands of zoos around the world and had a huge portfolio of animals allowing each customer to build their own unique offering. Even in the case of exotic animals like platypuses, ZPS had the scale that allowed it to build teams of specialists who knew everything there was to know about the animal. In the case of more common animals, ZPS did pioneering research and offered the gold standard in animal care. For example, lions cared for by ZPS lived longer and remained healthier compared to any other lion care provider. Using ZPS to care for your lions minimised the risk that events involving lions had to be cancelled and strengthened your animal care credentials.

As competition with Futurama heated up, Great Legacy started organising breeding for endangered species and started borrowing animals from other zoos for this purpose. The need for in-house expertise increased further but new specialists were too expensive to recruit. Furthermore, specialists wanted to work in teams with other colleagues specialising in the same field but Great Legacy needed just one specialist in each field and sometimes offered a part-time contract. It became increasingly difficult for the Great Legacy zoo to compete against Futurama on both quality and cost.

Introduction to PaaS offerings – the happy medium between SaaS and IaaS

By now, Software as a Service (SaaS) is an established business model in many industries. SaaS is the business and operational model by which a software supplier offers an application as a subscription service. The supplier arranges the housing, hosting, technical maintenance, security and compliance and all software licenses and updates to use the application. Although it would be ideal for an organisation if all of its applications were made available via the SaaS delivery model, it is highly unlikely this will ever be achieved across all industries and sectors. A SaaS supplier needs a minimum scale of users to operate and there are many applications that do not reach the critical mass to become affordable as SaaS. Furthermore, there are applications where tailor-made adaptations by customers, integration requirements or technical (legacy) constraints do not allow the delivery via SaaS.

In those cases an alternative to SaaS is the running of the application in the private or public cloud using the Infrastructure as a Service (IaaS) delivery model. Under IaaS, the customer outsources the housing and hosting of its application to a cloud provider. The cloud provider provides servers as a service in the form of virtual machines including secure access, backup and disaster recovery. The customer benefits by not having to organise data centre housing, setting up and maintenance of data centre hardware or setting up and maintenance of virtualisation software. All other activities, such as server software patching, monitoring, database management, load balancing, server security, and application maintenance, have to be organised by the customer.

In the course of the last decade, the public cloud has developed a set of services that fill the gap between IaaS and SaaS offerings. If we look at the IaaS infrastructure of an organisation in the private cloud, we find components such as database servers, load balancers, operating system software patching solutions, server monitoring solutions, file servers, active directory servers, data warehouses etc. All these components run on servers that need to be set up, monitored, backed up, patched regularly etc. The large public cloud providers offer these components as platform services (hence Platform as a Service). If you run an application server in the public cloud, you do not need to set up a database server, a data warehouse server, a separate server for antivirus protection, another server for monitoring, another server to manage software patches, another server or virtual appliance to have a load balancer. All these components/functions are offered to you as PaaS offerings by the public cloud provider.

What are the benefits of PaaS?

PaaS services are very similar to exotic animals in a zoo. Each service requires a lot of expertise and effort to set up and run based on IaaS services. PaaS removes all the hassle and complexity involved in the setting up and operation of technical components while increasing flexibility and transparency:

  • No setting up: you do not need to organise a project, find the engineers with the required expertise, purchase software, set-up the platform (including all required sub-components) and document the solution.
  • No maintenance: the public cloud provider manages the PaaS component. Backups, updates, monitoring and continuous improvements are all part of the service.
  • Rapid move from decision to implementation: PaaS services are ready to go out of the box.
  • No compliance risk on all required software licenses during the lifetime of the service.
  • Transparent pricing: no need to build a TCO model for charge through to the business because the pricing is based on a straightforward pay-as-you-use model with no separate sub-components, complex licenses and investments.
  • Extremely high SLA out of the box.
  • No security compliance risks during the lifetime of the service.
  • Flexibility: given the low entry and exit barriers, it is very easy to scale up and down the use of a PaaS service.
  • Pay as you go: you only pay for what you are using.
  • Peace of mind: ensuring high availability and smooth operation is very difficult in the IaaS model.

Moving technical components from IaaS to PaaS has a huge positive impact on complexity, reliability, flexibility, security and cost reduction. In the example mentioned in the introduction, the company moved to AWS and all 65 infrastructural virtual machines moved to PaaS. As the large public cloud providers keep adding PaaS services, the chasm between the public and private cloud will only grow further. It is quite obvious why Futurama selected Zoo Platform Services to support it.

Our advice for IaaS, PaaS or SaaS?

First, map your IT application landscape against SaaS services and identify which applications can move to this delivery model. The remaining applications have to be run on IaaS or PaaS platforms.

In order to select the best provider for your IaaS and PaaS needs, make sure you understand what PaaS services are offered by AWS, Azure and GCP. The breadth and depth of PaaS services varies greatly across the three hyperscale providers. There is not a single public cloud market but individual providers with differentiated offerings, priorities and cultures.

Once you understand what each provider has to offer, determine which provider best fits your current and future needs. Do not underestimate the importance of future needs when selecting a provider; optimising on the very short term rarely leads to the best TCO or IT effectiveness over the mid term. Another important criterion is to what extent a provider allows you to minimise the use of IaaS by deploying as many components as possible in PaaS.

Finally, select a single hyperscale provider to run your IaaS and PaaS services; spreading systems across multiple providers increases complexity and is rarely beneficial. We shall post a review of the main three hyperscale providers in another post.

—————-

Dionysis Linardatos

Co-founder and CTO of Cambrian Technologies

Dionysis brings twenty years of experience in IT and infrastructure development. Before starting Cambrian Technologies, Dionysis founded Sycada Mobile Solutions and he managed IT departments at multiple large corporations and start-ups.

Feel free to repost this blog on your website or social channels! But when you do so, please be so kind to mention the source and give us a notice via info@cambrian-technologies.com.