If we look back a few years, hybrid cloud was just beginning to enter the vernacular of IT departments. According to many industry analysts, hybrid cloud meant connecting an organization’s software-driven private cloud with the capacity on demand of public cloud, plus the orchestration, automation, and billing that allows workloads to move seamlessly between them. In reality, hybrid cloud was very seldom implemented this way. In this article, we’ll look at why companies struggled to bridge private and public, and why 2017 will be the year that yields a new, better definition of hybrid.
Why Hybrid Cloud Stalled
There were a few primary reasons that hybrid didn’t deliver the promised panacea. First, the infrastructure at AWS or Azure was different than what organizations had on-premises, making it difficult to move applications between them. And, in fact, in many ways it was hard (at best) or impossible (at worst) to even get workloads out of on-premises systems and into the public cloud without doing complete, clean re-installations. Other cloud providers that had a similar infrastructure design to on-premises data centers were also unable to deliver on the promise and expectation of public cloud in terms of services, pricing, and agility.
But, if you wanted to move stateful applications between clouds, you have to handle the persistent data, too. Moving data-rich applications meant either extensive downtime due to the long time to move large data, or untenable latency if data was kept on-premises, since most data-rich applications were designed with the assumption that their data is co-located with the applications.
In the event that you were able to reinstall your applications in the cloud and get the data sorted out properly, management was still problematic. It took roughly the same amount of time, energy, and resources to manage public cloud resources in parallel with an existing private cloud. So, in many cases, it was more challenging because you were basically managing two unique entities.
More recently, the definition of hybrid cloud shifted from the classic combination of a tightly tethered private and public cloud to what most enterprises consider it: Primary, on-premises data centers augmented with a growing combination of public cloud services like software-as-a-service (SaaS), platform-as-a-service (PaaS), and infrastructure-as-a-service (IaaS).
In fact, public cloud has been making huge strides recently — with providers like Amazon and Microsoft announcing substantial revenue growth. These providers are also making huge strides in their services: building out more regions, supporting more diverse hardware, improving security and compliance, reducing costs, and improving interoperability. Because of these shifts, Gartner estimated impressive growth for IaaS in 2016 — 42.8 percent year-over-year to be exact.
The writing is on the wall: the quality of available services combined with continuing price drops are making the lure of public cloud inescapable for most organizations. But for organizations that already have large datacenter investments, the question of how to build the right ‘hybrid’ model remains open. Below are the trends we believe will drive tremendous hybrid cloud growth in 2017.
First, we’ll see true workload mobility, whereby enterprises can seamlessly move application workloads between on-premises data centers and public cloud. In the past, enterprises have had to rely on technologies borrowed from backup and disaster recovery methodologies that were designed for largely static data sets. These replication-based migration methods require synchronizing entire data sets into the cloud before an application can even begin to run. In addition to that, once the data set and application workload began executing in the cloud, there typically wasn’t any easy way to get it back on-premises if needed. This meant that cloud migrations were time consuming, risky, and largely one-way, which essentially eliminates the ‘hybrid’ part of hybrid cloud.
New techniques (such as streaming-based migration) have already begun to gain significant traction, as they get application workloads to begin in the public cloud in just a few minutes, which vastly reduces wait-times for data transfers, but while maintaining performance, functionality, and uptime of the applications. These capabilities will give you the power to do significantly more pre-migration testing, as well as provide true capacity-on-demand, by adapting more dynamically to real-time shifts in computing requirements.
Second, we’ll see multi-vendor public cloud strategies start to proliferate. Workload mobility will begin to support easier private cloud rollback and/or cross-cloud-migration, which will improve your ability to mix and match your cloud architecture. You’ll be able to select the ideal cloud vendor, region, and instance-type based on characteristics such as cost, performance, functionality, availability, and proximity to end users. This will make one-way-only cloud migrations a thing of the past.
Third, we’ll see a more seamless bridge between private and public cloud. Independent software vendors, system integrators, and even cloud providers will increasingly rely on each other’s application program interfaces (APIs) to provide orchestration and management solutions, which will drive dramatic improvements across management. Furthermore, workload mobility solutions will continue to leverage APIs to create plug-ins that are integrated directly into existing management consoles, letting IT quickly manage cloud migrations from the same familiar interfaces (instead of a plethora of new consoles). We are, in fact, already seeing this trend across solutions like CSC Servicemesh, HP CSA, VMware VRA, and others. In addition to that, however, you’ll begin to sidestep consoles altogether in favor of automation tools for complex, multitiered applications that are too cumbersome to move via consoles.
Fourth, facing the great challenge of delays, which are caused by the application migration itself, of which there are three current strategies:
- Re-host, where you keep the application mostly as-is, but migrate it from on-premises and into the cloud at the platform level. This is sometimes referred to as, perhaps deceptively so, “lift and shift”.
- Re-factor (or Re-platform), where you make some modular changes to the application itself, for example swapping out SQL for Relational Database Service (RDS) before moving to the cloud.
- Re-write, where you (almost) completely recode an application for optimal cloud performance, which of course requires a substantial investment of time, energy, and money.
A few things are going to drastically change regarding how you approach these application challenges in the near future. To begin with, you will begin moving these applications to the cloud first so that you can enjoy the benefits of the cloud while you optimize the application in parallel. Additionally, solutions will begin to automate these optimizations and adaptations, removing friction from the migration process. As solutions make it easier to get applications in the cloud and automate optimization tasks, long application re-writes will become less necessary. This will empower you to simply re-platform or re-factor, greatly reducing your time, energy, and monetary investments, and powering significantly faster mass cloud migrations.
Lastly, cost reduction will become much more attainable through cloud workload mobility automation. These solutions will monitor workloads and analyze their requirements, while in parallel assessing the available options from cloud vendors, regions, and instance types. The cloud workload mobility solutions will be able to dynamically shift workloads across these components to best realize your top priorities. For example, if high performance is the top priority, workloads will shift into larger instance types that are close by when needed. Or, if low cost is the top priority, workloads can be automatically shifted into more cost-effective regions or cheaper, temporary instances to maximize cost savings. Data, of course, will remain synchronized and persistent throughout these dynamic architecture shifts.
As we’re about to see, 2017 is going to be a powerful year for both public cloud and enterprise hybrid cloud deployments. The bridges between the private and public cloud will continue to fuse into something that is a seamless, simple, and efficient enterprise cloud architecture, and you will begin to see the benefits grow more rapidly.