This is a guest post for the Computer Weekly Developer Network written by James Sturrock in his role as director of systems engineering at Nutanix.
Sturrock writes in full as follows…
In the modern era of cloud computing, our enterprise application success is defined by our ability to connect, integrate and forge interrelated bonds between essential and data services and their core functions alongside their ancillary dependencies and beyond.
Okay yes, enterprise application success is also essentially defined by our ability to adhere to regulatory compliance and governance legislature, to remain robust and secure in the face of cyber threats and the entire IT stack’s ability to provide user functionality – but the point is still valid – all of those elements of technology also rely on their ability to connect to our source SaaS service motherlode, to each other and to external services that may support their very existence.
My proposition to you (imagine we’re in ‘the house’, or some other debating forum if you wish) is that all of these parts of the software development and delivery cycle depend upon a cloud platform, a cloud foundation and from a precision-engineered cloud infrastructure.
Initialise with infrastructure
By adopting an infrastructure-first approach, organisations can develop SaaS connectivity and integration services based upon a foundation that is capable of unifying a hybrid cloud infrastructure made up of compute, storage, networking, hypervisors and containers.
Self-healing autonomous controls for disaster recovery, resilience provisioning and security should never be thought of or considered as a bolt-on afterthought, these are the core attributes of cloud that a business should be adopting as built-in functions on the platform of choice that they rely upon.
Building cloud layers
Like any integration strategy, a prudent approach to cloud infrastructure starts with a core set of software functionality services for essential system resilience and management.
That’s fine as an opening statement (stay with me on the debate motif if you can), but what kinds of functions should an enterprise be able to draw upon for SaaS integration to be made as fluid as possible within the infrastructure services it relies upon?
Within the core functions an organisation should be looking for if it wants to run on a cloud infrastructure capable of working, flexing and integrating in the way we have always been promised by the cloud evangelists, a business needs to be able to react to the complexity of modern business by being able to run heterogeneous cloud clusters.
Integrating across heterogeneous instances is essential if firms are going to be able to work with complex supply chains that may switch focus in days, weeks and sometimes hours. It also needs to be able to access data tiering, tunable redundancy functionality and core services such as data deduplication i.e. there is much to swallow here and this is why it’s an infrastructure move first and foremost.
If I may point to our own Nutanix Cloud Manager (NCM) service here, this is an exemplification of the types of functionality we have proposed so far. This technology offers simplicity and ease of use to build and grow a cloud deployment faster and that naturally leads to firms being able to rapidly realise ROI with intelligent operations, self-service and orchestration, visibility and governance of spend, security and that’s all through a unified multi-cloud management solution.
But away from our own platform, these factors tabled thus far should underline why the march (we could be more corporate and say ‘mission’ if you wish) towards integration all starts with infrastructure.
Think about the use of Infrastructure-as-Code (IaC) now helping organisations to leapfrog the misconfiguration headaches so often associated with cloud deployments perhaps a decade ago.
Think about the rise of serverless technology and the software-defined world of networking interconnectivity that we now champion as a core part of what it takes to deploy modern applications and data services over increasingly cloud-native frameworks.
Think about the need to grasp multi-cloud choice across Cloud Services Provider (CSP) hyperscalers, across an organisation’s choice of hypervisor, across its preferred container platform and throughout the landscape of services that it uses to graph data, storage and analytics services. You might even think about the need to engineer for GPU support now that we see huge new factors of acceleration and automation stemming from the cloud.
These are all validation points for infrastructure first, integration second – and that can be immediately afterwards when and where needed.
Where all this gets us to – logically enough to round out our debate proposition – is the need for cloud management technology. With the reality of multi-cloud across hybrid estates and IT stacks, teams now need to orchestrate intelligent operations and take advantage of self-service technologies with a view to gaining visibility, observability and enabling governance in unified environments where incongruent incompatibilities are rare, if not completely absent.
If organisations adopt this infrastructure-first approach to cloud computing, then the SaaS connectivity and SaaS integration aspect of any firm’s IT deployment can be engineered for the continuous world of always-on computing.
Starting with the front end and thinking about infrastructure considerations is akin to a business building a shop window but failing to stock its shelves with any products, let alone thinking about how it will handle its supply chain and wider network of customers and third-party partners.
SaaS integration starts with infrastructure and ends with it as well – build your base, then build your business.