The Journey to Cloud Article 4 PaaS and SaaS (and Managed IaaS)

IMG_0059.jpg

We have now transitioned ourselves from an owned hardware model to a consumption-based commodity model - but this is only the starting line.

In the process of getting here we have consolidated and reduced our hardware and software (operating systems and applications) but we may have increased our number of deployed servers.

Let’s look back at the example we used in virtualisation.

See The Journey to Cloud Article 2: Prerequisites for Moving to as-a-Service

We still have a massive server/database/application deployment but it is now better managed and on a usage base:

  • 16 different operating systems – reduced based on less hardware platforms and upgrading OS to supported versions

  • 6 different databases running 10 different versions

  • 1000+ different applications, most of which we did not develop, just purchased

  • 10+ different java/.NET versions and clients

  • 200 different web portals, thin and thick clients

Now the real work begins

We need to transform the way we develop, manage, and support business outcomes to really take advantage of the services offered by the Cloud Provider and begin moving up the stack, understanding the Cloud functions and planning how to get the best outcomes from them.

The first step is leveraging the Database as a Service (DBaaS) offering of the Cloud Provider. This can drive significant savings and simplification.

We should be able to move to 2 databases at a max, running 1 version each:

For example, Microsoft Azure Standard SQL is ~$223.20USD per 100 GB per month, and has 99.99% availability.

 Again, the decision for which Database as a Service on which Provider should be a cost, Service Level Agreement (SLA), and feature analysis.

We still have a lot of servers but we have simplified our architecture and handed over our databases to an expert in the database. Now we just consume database services - this allows development to focus on outcomes and data analytics rather than the care and feeding of the underlying database systems.

Additional services we can use

 So let’s keep going with other services the Cloud Provider offers:

  • Load balancing services

  • Content delivery networks (CDN) and Gateways

  • Authentication and Identity Services (Including Microsoft AD as a Service)

  • DNS services

  • System monitoring and management services, including security assessment and firewall services

  • Certificate Management Services (CMS)

  • Analytics services - Machine learning and AI services

  • Message queuing services

  • Desktop as a Service (DaaS)

At this point the number of real servers we have has been significantly reduced, and bespoke or custom build applications and functions should be reduced to a minimum.

We still have applications deployed on servers we manage:

  • 4 different operating systems – reduced based on less requirement and standardisation. All OS are vendor supported and subscribed to through a Cloud marketplace

  • 2 Database as a Service solutions, SQL and noSQL

  • <1000 different applications

  • 5 -10 different java/.NET versions and clients – reduced based on simplification of services consumed

  • <100 different web portals, thin and thick clients

We are almost there

Now we move the remaining virtual machines to managed Infrastructure as a Service (IaaS) and Platform as a Service (PaaS), handing support of the operating systems over to the Cloud Provider and vendor ecosystems that supply these services on the Cloud providers. Now we can focus on development and running the applications on the systems ourselves, depending on how standard the applications are.

Standard Linux web servers, accounting programs and email services can easily be moved to PaaS:

  • Email services

  • Accounting services

  • HR services

  • Web server, shopping cart, and credit card services

  • Printing services

  • Sales tracking services

  • Inventory management services

  • Order tracking services

All of the above can be moved to PaaS and integrated with the existing DBaaS implementation and Storage Services we are already using. We can use our machine learning and AI across the information stored in these PaaS applications.

We still have middleware and processing applications deployed on up to 100 servers where we only manage the applications:

  • All the servers operating systems are managed by the Service Provider

  • 2 Database as a Service solutions, SQL and noSQL

  • <500 different applications

  • 2 - 4 different java/.NET versions and clients – reduced based on simplification of services consumed

  • <50 different web portals, thin and thick clients

Now that we are truly transforming how we develop capability and business outcomes continue the journey in the next article ‘Serverless’