Cloud computation is a hot topic across industries that require scalable software solutions to enable growth, and for good reason. The cloud, in all its mystical appeal, has so mellifluously penetrated the zeitgeist that experts engaging with new software solutions are often eager to hear how they fit into the cloud computing ecosystem. In the world of next-generation sequencing (NGS), where potentially massive data throughput necessitates substantial computational resources, the appeal of cloud deployment is exacerbated. However, there is still value in doing things the old-fashioned way with a local deployment of key software infrastructure. The upshot is that the Golden Helix software suite supports both private cloud and local deployment, and there are plenty of good reasons to consider either or both, depending on your use case and how it evolves over time.

Let us start with some important background information. A core pillar of our software solution is that our users own all of their data, and we do not collect nor store any NGS data that users upload into the software. This is true in both a local deployment setting and when the software is deployed on the cloud. Users are free to (and encouraged to) export and back up data processed with VarSeq at their own chosen cadence. While this is an enticing data security topic in its own right, it also means, from a deployment perspective, that users’ analysis is not tied to a specific machine or environment. Hence, users are afforded the nimbleness of being able to replicate their production infrastructure (input data, analysis pipeline, annotations, etc.) in a different environment as needed. Thus, users can switch between local and cloud deployments as needs and costs evolve.

So, given that users can not only deploy locally or in the cloud but can also port production environments and data between the two paradigms, what factors should be considered when choosing or changing a deployment strategy? Bearing in mind that the deployment strategies for local and cloud infrastructure can be further divided into many options with different pros and cons, let’s outline the main considerations at the most basic level. We can look at both options regarding ease of use, cost, and scalability.

In terms of usability, local deployments benefit from the fact that users can define every component of the environment, from the operating system and other software choices to hardware specifications like network speed, number of available CPUs, and RAM, which affords more customization options at the cost of potentially being more difficult to set up. Conversely, while it can certainly be easier to set up a private cloud instance and requires no physical labor setting up servers and network storage, there are still plenty of pitfalls when getting started with a cloud deployment, and users tend to have a narrower set of options for things like computational resources and support software. In addition, institutions will typically still need a team dedicated to maintaining their cloud environment, similar to what they’d require with a local deployment. From the end-user perspective, however, especially when it comes to interfacing with VarSeq and Golden Helix’s other software solutions, the experience can be virtually identical.

When it comes to cost, numbers will vary vastly between groups depending on expected throughput, staffing, and other considerations such as existing infrastructure. The main consistent difference is that most of the cost of a local deployment will be upfront, whereas the cloud tends to be billed regularly depending on usage or other pricing models. A situation with predictable throughput requirements can make a local deployment an attractive choice with well-defined expected costs.

Regarding scalability, you can’t beat the versatility of scaling with cloud deployments. Whereas a local deployment requires additional physical hardware to be acquired and installed to increase computational capacity, cloud deployments can often be augmented with additional capacity at the click of a button. Similarly, when users’ needs change, and less throughput is required, cloud deployments can be scaled back appropriately, while physical hardware obviously cannot. This makes the cloud a good choice when throughput is highly variable or unpredictable. However, our support and development teams do everything we can to enable users to dial in their analysis pipelines and improve throughput steadily over time, which is an appropriate use case for a local deployment.

In summary, there are plenty of pros and cons to both cloud and local deployments for analysis with Golden Helix’s software suite, and we’re proud to be able to support our users regardless of what they choose. Ultimately, users should weigh the costs of both options with their expected throughput and design a deployment strategy that meets their needs. If you are interested in discussing VarSeq for your clinical analysis, please visit us online and request a free trial here. Also, for current Golden Helix users, please don’t hesitate to contact [email protected]m if you want to review your options.