There has been a rash of stories in the IT press lately regarding how hybrid cloud is going to be the predominant model for adoption of cloud services among organisations, with the implication seeming to be that this is some shock revelation that nobody expected.
Cloud computing is still a relatively young sector of the IT industry, with Amazon’s S3 service launching back in 2006 and a handful of applications such as webmail pre-dating that. For that reason alone, it is unlikely to have yet reached the point where it has taken over the business computing market, and especially when you consider that large enterprises tend towards being conservative and moving somewhat cautiously.
Nevertheless, cloud computing has grown swiftly, and uptake continues to expand. There are good reasons for this; services such as those from Amazon’s AWS stable have demonstrated that some processes can be offloaded to the cloud with a cost saving to the customer. Email is a good example, as is software development and testing, along with backup and disaster recovery.
Cost savings typically come from the customer not having to purchase, deploy and maintain their own on-premise infrastructure in order to operate the service. Another key factor is flexibility, with cloud services holding out the promise of customers being able to pay for only those resources they use, and even swap service providers or bring service provisioning back internally if they so desire without having to pay a penalty charge. However, the reality often doesn’t quite add up.
Hybrid cloud gets the name from the fact that it is a combination of on-premise IT service delivery and some IT services delivered from an external provider, often a public cloud platform such as AWS. In the first case, the organisation may have its own “private cloud” operating on its own infrastructure, but this is not always the case.
Having a “private cloud” implies that the organisation has organised its internal IT infrastructure into a giant pool of resources that can then be carved up as necessary in order to meet the demands of various user groups and applications. Often this is delivered by putting in place some sort of self-service kiosk or portal user interface that allows users to request the compute resources they need. In other words, it needs to act like a public cloud service, such as AWS.
However, this requires a high degree of automation at the management layer in order to be able to provision resources on demand, and to be able to keep track of used and available resources. It also implies that the infrastructure itself must be capable of being carved up in this manner, which traditional hardware was not really designed for. For this reason, many enterprise firms still have a lot of legacy IT kit that works perfectly well, but may not be easily co-opted into an on-premise private cloud.
But this does not stop organisations from taking advantage of cloud services. In fact, many will have started out on the road to cloud adoption through their development teams turning to AWS for speedy procurement of servers and other resources for initial development and test work, rather than go through the often interminable process of getting appropriate resources from their own IT department.
Later, those organisations may well have decided that it might be wiser to bring their development hosting back behind the firewall, which would entail building something on-premises that functioned like AWS. Often, this would require new-build infrastructure, and IT vendors quickly sensed an opportunity and started to offer turnkey solutions designed to do just that; converged infrastructure composed of servers, storage and networking kit all pre-configured to work together, or more recently, hyperconverged systems which integrate server and storage together into an appliance-like node, ideal for scaling out as demand requires.
Using cloud for greater flexibility
But even with an on-premises private cloud, organisations may find that they still have a requirement for public cloud services. The classic example is to meet seasonal demand, such as an online shopping site needing extra resources in order to cope with a peak in the volume of orders in the run-up to the Christmas period.
This is the source of the term cloud bursting, meaning that a business’s compute capacity can “burst” out beyond the bounds of the organisation’s own infrastructure in order to continue to service an increase in workload, without the business having to own enough infrastructure to meet these peaks in demand, resources that would be underused for the rest of the time.
That is just one scenario for how an organisation could arrive at a hybrid cloud deployment, but there are others, of course. For example, an enterprise may prefer to host its public-facing website in a public cloud provider’s datacentre, in order to keep it separate from the on-premise IT infrastructure it uses to actually operate the business.
And so far, this discussion has largely focused on infrastructure services delivered from a cloud, otherwise known as infrastructure as a service (IaaS). Others available include platform as a service (PaaS), which typically refers to a hosted environment for developing and running applications; and software as a service (SaaS), which refers to cloud-hosted applications themselves, such as email or CRM.
Often, the services that any given enterprise customer requires are likely to come from more than one provider, so that an organisation may decide that its app development needs may be best met by AWS, while email may be outsourced to Microsoft’s Office 365 service and off-site disaster recovery entrusted to a third cloud provider. In fact, some recent surveys have shown that many enterprises are using or piloting services from as many as six different providers.
This arrangement is described as a multi-cloud strategy, yet another industry buzzword, and can also be understood as a way for organisations to hedge their bets against getting locked into one cloud provider.
Cloud computing may be still a relatively young sector of the IT industry, but it has already come a long way, and is now accepted as just another tool in the IT department kit for meeting the compute requirements of their organisation. There is now a broad range of cloud services available on the market, offering different capabilities and designed to meet a range of different requirements.
This is just as well, as it is unlikely that any single cloud provider is likely to meet all the requirements of an enterprise customer. It is also unlikely that enterprises are going to entrust all of their IT requirements to a cloud service provider now or for the foreseeable future.
You can expect the majority of organisations to keep their most critical IT functions operating internally, and when you realise that, it is pretty obvious that hybrid cloud will be the predominant model of cloud computing for some time to come.