Keeping musedata in the cloud

Big data and analytics go hand in hand with embracing the cloud; the two together form half the digital ‘SMAC stack’ in the museum (referring to social, mobile, analytics and cloud computing). Given the sheer scale and exponential growth of data required behind analytics solutions, the cloud provides crucial business benefits for analytics outcomes. However, from a governance perspective in cultural organizations, the cloud can represent a new unknown, and therefore be perceived as inherently risky. In reality, cloud computing helps to address many of these same concerns such as security and stability which can be unrealized characteristics of the museum’s current technology landscape. Analytics provides a great opportunity to advance the cloud cause without touching highly critical processes or sensitive data functions, to effectively experiment with a cloud future.

As a quick overview, the cloud essentially refers to remote computing resource accessed via the Internet, commonly via a browser, as a utility. This offers a great deal of business benefit over systems physically stored on site or in a data center, also known as on premise, or more fondly, the ‘broom closet’, given many servers are kept in less than ideal environments rather than a managed data center.

Cloud offerings generally fall into one of three levels: Infrastructure as a Service (IaaS) – a bare computing resource controlled from the platform up; Platform as a Service (PaaS) – a built environment controlled from the application up and Software as a Service (SaaS) – a complete software controlled at the data layer only. All levels are relevant in the realm of museum analytics, though a readily available and entry level solution will always begin at the software layer. Cloud is also generally classified as public (shared resource, though your particular tenancy and data remain private, visible only to you), private (dedicated resource), or community based (such as an all of government service). Across the musetech landscape, many cultural organizations find the best strategy across their mix of applications keeps ‘one foot on the ground, one in the cloud’ in a hybrid approach and may use a mix of tenancy approaches depending on the sensitivity and criticality of data involved. For example, whilst new analytics solutions find natural fit in the cloud, the comparatively more static world of collections management data may still be comfortable on premise, especially if migrating would require additional investment.

The benefits of cloud computing are not simply limited to cost in both initial outlay and overhead, though a significant advantage. As the cloud can often be provisioned almost instantly, without the lead time of server procurement, this ability to scale up (or down) as needed provides elasticity to help keep the cultural organization agile, which is especially important given analytics storage requirements can grow exponentially. Cost efficiencies are also catered for with a dynamic approach, whereby multiple tenants share cloud resource as opposed to needing to procure an entire server only to leverage a small portion of it and only for a certain amount of time on a regular basis. Take a museum analytics solution for example, where ingestion will be only so often each day and usage limited to working hours, with select spikes for processing insight. In many ways, the cloud is a manner of successful outsourcing, given the activities of managing infrastructure do not form part of the museum’s core business activities. Importantly, the cost to serve these solutions via the cloud is likely to be more closely linked with the organization’s usage, beneficial with increased financial pressures and transparency requirements.

In addition to its elastic and dynamic nature, cloud computing offers ubiquity, by being available anytime, anywhere on any device (including mobility, if supported at the software layer). This supports museums wanting to pursue remote working, bring your own device and external agency collaboration. From an analytics perspective, these benefits are important for leveraging the museum’s data outside of its walls, supporting conversations with governments, donors and the community, as well as freeing staff from behind the desk by allowing analytics reference out in the gallery or collaborative spaces.

Often, many cloud opponents cite risk and associated costs when appealing against a transition to the cloud. With the exception of extraordinarily critical or sensitive unique data points, a cultural institution would be hard pressed to say it is better positioned in the business of data risk management than a dedicated global hosting provider. A more likely scenario is that the broom closet server (as data storage of any kind brings with it a level of risk) is not environmentally isolated, geographically redundant or patch protected to the same degree as the specialist provider’s systems and most certainly will not have nearly the same level of security investment as these global players, offering less business continuity and disaster recovery (in terms of time to recover and how recent any recovery will be) in case of an event. That being said, the cloud is still vulnerable to risk – security is a responsibility at all levels of the cloud pyramid – a move to the cloud does not absolve a museum from managing technology risk governance. This includes evaluation of the cloud provider’s prevention, monitoring and response mechanisms, both system and procedural.

Another misunderstood return on investment is maintenance, a hidden cost. The initial capital outlay behind an on premise system is only a portion of the life cycle cost – often operational, corrective and preventative maintenance runs several multipliers of initial expenditure. Behind this sits a significant departure from upgrade based release cycles, which often require a project approach, introducing a migration load for any analytics data store. Instead, public cloud offerings usually leverage feature based releases where the core solution is quietly upgraded over time. This is especially important where analytics solutions support integrations, as these would otherwise require significant effort to transform. Maintenance concerns don’t just add cost and complexity to an on premise system but down time too, another misunderstood cloud aspect for those critical of availability, given upgrade migrations will likely take hours or days to complete.

However, one trade off for these economic benefits is a limitation in control and flexibility. As with any productized solution, feature and functionality fit will require review, along with roadmap alignment. Whilst some customizations can be supported through plugin style add ons in a cloud environment, intensive bespoke development may require a private instance, offsetting some capital and maintenance savings.

A determining factor in the success of any cloud solution is connectivity. On the system side, this is dependent on resource availability and solution uptime; on the client side this translates into the museum’s network connection speed, stability and data plan. Depending on how involved the user experience is, latency (any elapsed time between a client request and provider response) could also be an issue. Whilst this addressable concern may be critical for highly involved processes performed in volumes, such as complex ticketing, it is unlikely to be noticeable in the field of analytics.

Finally, regulatory oversight may dictate the conditions of cloud consumption for some cultural institutions, especially those that are government owned or publically funded. Often, one of the chief concerns pertains to data sovereignty and the subject of data to the regulatory oversight of a foreign nation, depending on the provider’s location or that of its data centres. For non personally identifiable analytic data, this concern should be limited to clarity on the provider’s data ownership, privacy and persistence policies. If the museum or other agencies under the concerning jurisdiction are already using a solution such as Google Analytics or a social media channel, the precedent for managing analytics in the cloud is already established. Where data begins to involve personally identifiable visitor data, such as a customer relationship management or member loyalty program, this situation may require review, along with compliance requirements especially if billing information is retained.