Hybrid and multi-cloud: economically, often the best choice
December 4 2019
by Owen Rogers
As it becomes an explicit IT strategy (and one now recognized even by hyperscalers, with their cloud-to-ground initiatives), hybrid cloud presents a major opportunity for service providers and technology vendors.
The 451 Take
Reducing lock-in for customers through multiple venues might appear risky – what if they want to leave? But far from a threat, hybrid and multi-cloud must be embraced as an opportunity by the tech industry; creating lock-in or reducing options might just alienate customers. Tech industry players should enable options but build out a preferred venue, so it is attractive through bundling capabilities, guarantees and a reasonable price. Once a provider is a trusted partner, most decision-makers, especially senior-level, are happy to use a preferred venue even if it is more expensive. To make this happen, new levels of abstraction are required, as are new levels of integration and management. Services that are composable and software-defined (i.e., cloud-native) will help. The IT landscape of the next decade is all about diversity – any infrastructure, any cloud, any app. It's not chaos, but opportunity; mastering multiple clouds is key.
Economics of hybrid and multi-cloud
Hybrid cloud – the combination of public and private cloud – is the new normal. A hybrid cloud approach is being adopted by 57% of enterprises, according to our research, primarily to enable developers and lines of business to select the optimal technology for each workload's specific requirements, the aim being to add value to the business. Cost is the second-most-cited reason for using hybrid cloud and, in principle, should be the most cost-effective means of providing scalable cloud consumption, provided the private cloud is being managed at high levels of labor efficiency and utilization, and the public cloud and private cloud are unified as a single cloud platform.
Specifically, utilization of the private cloud should be above 70% over the lifetime of the platform, but negotiating a discount of 30% off list significantly improves this breakeven to levels achievable by most enterprises. Furthermore, enterprises can deploy tactics to increase their utilization, such as using flexible hardware procurement and consumption models to avoid provisioning unused capacity up front. Aggregating applications from multiple departments and geographies can significantly improve utilization. In fact, the more workloads aggregated, the more likely a flat (and thus, more predictable) demand can be achieved, if these workloads' demand requirements are unrelated to each other.
A crucial factor for the value of a hybrid cloud is how the venues operate – unifying private and public cloud capacity saves 10% on enterprise direct spend compared with running each platform separately. Enterprises also consuming private cloud in an opex manner save 22% compared with those running separately managed public and private cloud platforms using a capex approach. Cloud-native frameworks are a crucial component in delivering a unified approach by taking advantage of the range of IT venues while moving attention away from infrastructure and onto the application. In this situation, applications can pull infrastructure dynamically to fulfill their needs, enabling true pay-per-use (rather than pay-per-provision) compute efficiency.
Using multiple public cloud providers can make direct savings, too. US customers can save half their cloud expenditure by mixing and matching services from multiple providers. If enterprises are willing to make up-front commitments, they can save 65% off list pricing. But total savings depend on how feasible it is to manage these cloud platforms, and there are challenges related to skills and management.
Continuous workload migration is a key use case for hybrid cloud – applications and workloads must be able to traverse on- and off-premises environments and be available anywhere at any time. Among businesses building their own software, 75% indicate that at least some is cloud-native, and that the ability to move applications between environments without significant refactoring is important (46% consider it very important). Such migration won't take place ad hoc or in response to second-by-second changes, but will be there to provide options should migration be required.