Connect with us

Brand Voice

Analytics on cloud platforms no longer a ‘nice to have’

Avatar

Published

on

By Kelly Lu, Specialist in Advanced and AI Practice, SAS Africa

The Covid-19 pandemic is likely to be remembered as the biggest driver of digital transformation in our lifetime. There has been a shift towards cloud adoption driven by the need of digital transformation, where more experimentation with data, scalable and flexible workloads and the introduction of new business models are the key reasons for organisations to migrate to the cloud.

Now, rather than simply being on the ‘to-do’ list for any IT department, cloud has now become the top priority, since businesses need to change their process and behaviour in order to continue operation.

It’s been a while coming

Previously, companies migrated to the cloud largely for IT-driven technical reasons, such as end-of-life hardware with expensive maintenance contracts, or a pure cost perspective. Owning hardware and infrastructure is simply not appealing any more, and organisations are moving from a CapEx to an OpEx model where a pay-as-you-go model is simply more sensible. It’s understandable that companies have reservations about the shift from on-premise to cloud-based services – analytics is a little late to the cloud party, for several reasons.

There is the concept of Data Gravity, which means that your workloads are pulled towards the location where the most heavy data resides. As long as the data is in your on-premise data centre, it’s also most logical that your analytics workloads are executed there.

Organisations are also not settling on one cloud service only – they are looking around and trying out different cloud vendor offerings and adopting a mix of public cloud services as well as Software as a Service (SaaS) applications that they are consuming from the cloud. Those SaaS applications often come with some form of embedded business intelligence and reporting. They may not fulfil all the business requirements and are built to support a very generic usage. However, they may be sufficient to get started and only at a later stage will there be additional requirements in the business, along with a need to start combining data over the multiple clouds and from those SaaS applications.

The market has expectations

Organisations have a number of expectations for cloud adoption. Other than reduced costs, agility is another key expectation, where the goal is faster time to market. To accelerate time to market, organisations are setting up DevOps processes that automate the deployment of applications and analytical models continuously. As soon as an organisation designs an analytical model, it has to enter an automated DevOps pipeline, undergo steps of testing, validation and versioning. Once all these tests are passed, it should come online and be published immediately.

Organisations also have high expectations in terms of performance and reliability. As workloads can be highly varied over time, the infrastructure needs to scale accordingly.

Security and compliance is no longer an excuse not to move to the cloud, with many cloud providers recognised to have higher standards when compared to some owned, on-premise data centres. Organisations also see many new use cases where the cloud allows them to experiment with new technology in exciting areas such as AI, IoT and of course, digital transformation. Cloud enables companies to analyse enormous amounts of data and embed analytics into their decision-making processes. The pandemic has driven new sources of data and new customer behaviour that doesn’t rest on a lot of historical data, so it requires new analysis, which cloud services can provide faster and with more agility.

Where to start

It may seem daunting to know where to get started, but there are ample best practices available. These provide information on tested cloud infrastructure and benchmarks, as well as typical steps for not only deploying software on cloud infrastructure, but also what to look out for when migrating company data and applications to the cloud. Business continuity is also a key consideration, as no organisation wants to interrupt their operations while migrating. There are best practices for this too, usually with a window period involving parallel running to ensure that the platform is stable and integrated with existing services.

One scenario that many companies may have heard of is ‘Lift & Shift’. The idea is to replicate an existing on-premises environment in the cloud without changing anything to it unless absolutely necessary. Public vendors in the cloud industry even offer tools to make a ‘Lift & Shift’ transition easier. This approach is very appealing to organisations, particularly if the goal is to be up and running in the cloud quickly. However, the journey should not stop there, since ‘Lift & Shift’ does not provide the full benefits of the cloud.  

A clear cloud adoption strategy is crucial

It’s important for organisations to have a clear cloud adoption strategy that will help choose the best migration approach besides ‘Lift & Shift’. There are alternatives such as rearchitecting or completely refactoring your application when moving it to the cloud. These approaches have the goal of integrating deeper into the services that cloud providers offer to benefit from the high availability and scalability that these can offer, often at a lower cost than organisations would find if putting all these services together themselves. Here it’s important for organisations to have a clear understanding of their workload to get the most from their target architecture. There is a tremendous amount of brilliant technology available at organisations’ fingertips to help operationalise and automate models while providing benefits such as reduced costs, unlimited access to storage and easy scalability. The cloud also enables experimentation to innovate and scale analytics quickly and precisely.

A completely new architecture for SAS Viya

SAS Viya is designed as cloud-native, which means it is built from a number of components, all loosely coupled, independent and scalable. These components are deployed inside containers and can be updated using continuous delivery mechanisms. New versions of Viya are available every month, which allows organisations to continuously benefit from the latest features and functions without having to go through heavy upgrade and migration projects.

Viya is designed as Open, which helps integrate with cloud vendors’ native services, for example authentication, for utilising databases and storage, and also for orchestrating all the containers. For orchestration of containers there is a market standard available, Kubernetes. All the public cloud providers offer Kubernetes as a service, and organisations can also choose to deploy Kubernetes in their own data centre if they prefer to deploy the latest SAS Viya release on premises.

Advertisement