Load balancing object storage in hospitals

The use case for object storage in hospitals, and why it needs a dedicated load balancing solution

Healthcare data storage solutions need to be able to serve users from different locations and provide access to large quantities of data quickly and easily. The additional challenge faced by hospitals is the unprecedented legacy spaghetti of complex, existing data systems. Couple this with the serious consequences of data loss or downtime, and it is understandable why hospitals have traditionally taken a more risk-averse approach to IT infrastructure modernization. But has the time now come for object storage to be embraced more widely across healthcare, and applied to a broader range of use cases?

What is object storage?

Object storage is essentially a large, metaphorical bucket where you can dump huge amounts of unstructured data. Each piece of data is carved up and chronicled using a unique metadata description that allows it to be easily identified and retrieved at a later date when needed.

Let's borrow the excellent analogy of a cloakroom attendant, used by one of our object storage partners...

Your unstructured data or 'object' is akin to a coat you leave with the cloakroom attendant at the beginning of a night out. Later on in the evening, you hand over the tag you were given at drop-off which was unique to your garment, and the cloakroom attendant then uses that information (read 'metadata'), which then allows your specific coat to be easily retrieved from the correct clothes rail. As such, we engage with object storage in a completely different way from traditional data files.

When might hospitals use object storage instead of file storage?

By breaking down unstructured data into smaller metadata, much bigger blocks of data can eventually be retrieved once they've been identified. File storage, on the other hand, stores data in structured, hierarchical files within folders, and transfers the entire data set in one go, making it much less proficient for large amounts of data.

Within the context of a hospital, this makes file storage much harder to scale, as it requires systems to be continuously added to meet the exponential increase in patient data demand. On the other hand, because object storage is kept in a single repository, and because the metadata can be extremely detailed, it can be found, managed, or analyzed much more easily. As such, hospitals are increasingly turning to object storage to solve a wider number of healthcare data challenges.

Use cases for object storage in hospitals

Here are just a handful of scenarios where object storage in hospitals might make sense:

• Medical imaging - archiving vast numbers of large medical imaging files for consultation or research. This example is explored in more detail at the end of this article.
• Immutable backups and ransomware protection – Often healthcare organizations and hospitals are the targets of devastating ransomware attacks so (with its ability to store incorruptible, immutable data backups) object storage is a perfect weapon in a hospital's cybersecurity arsenal.
• Scalable data storage – With object storage, files are stored as 'objects' rather than files on a local file system. This layer of separation between the offloaded low-level files, keeps object data table lookups at a more manageable size, making it possible to scale to hundreds of petabytes without experiencing degraded performance.
• Digitalization of patient records – A huge backlog of paperwork still exists in many hospitals, needing to be scanned and digitalized. Object storage allows data to be picked up and securely transported across the internet to wherever it's needed, making it easily accessible to users in any location.

• Medical research – During the pandemic, AstraZeneca used petabytes of genomic sequencing data to inform their drug research and accelerate their covid vaccine development, crunching massive amounts of data in a very short time using object storage.
• Data lakes – Pooling data into a vast data lake is also an increasingly common use case, with artificial intelligence applied to that data to help improve workflows, spot patient trends, and improve clinical judgments.
• Clinical decision-making support – Because the metadata includes lots of rich information, searching for a group of patients over 50 that presented a specific condition and running analysis on that dataset is made easier at scale. In this way object storage allows artificial intelligence and machine learning to be more extensively exploited.
• Medical imaging - see below.

Why object storage needs load balancing

While object storage offers some amazing stand-alone solutions, to achieve the full benefits of object storage, the application needs to be load balanced.

In a nutshell, load balancers maximize the benefits of object storage by providing:

• Highly available object storage – Having a sophisticated storage solution without accounting for proper high availability may not result in optimal performance. Therefore, by adding a load balancer, downtime for the end-user is avoided.
• Resilient object storage – Dedicated load balancing solutions provide a turnkey solution for object storage applications, providing immediate failover in the event of a storage node failure, providing a seamless user experience.
• Scalable object storage – Load balancing provides fast, redundant, scalable architecture for the transfer, storage, and retrieval of data. Infrastructure can be scaled up and down, mitigating the need for constant DNS and firewall configuration changes. Servers can be brought on and offline at the touch of a button, making change controls simpler. And optimal performance is ensured, as the load balancer conducts regular health checks and spreads network traffic across multiple servers as needed.

Why object storage needs a dedicated load balancing solution

An object storage provider would be crazy not to use a load balancer to help them deliver their SLAs (Service Level Agreements). A dedicated, clustered pair of load balancers avoids what we call a 'noisy neighbor situation', where resources are drained by other applications, which are then pulled away from the object storage application, impacting its performance.

In this way, dedicated load balancing guarantees intelligent load balancing capabilities to route and reroute traffic in a fraction of a second to deliver exceptional user experiences. The load balancing solution can be tuned specifically to ensure the high availability of the object storage application, without having to take the needs of other applications into account. Unsurprisingly this per-app approach is always recommended for critical applications, in order to ensure zero downtime.

Let's take medical imaging as an example.

Use case deep-dive: Load balancing object storage used to store medical images

One of the most convincing use cases for load-balancing object storage in hospitals is to facilitate the storage, analysis, and recovery of archived medical images. Load balancing these files ensures these images are always available to those who need them, wherever they are. This supports more agile working practices, such as clinicians working from home, or the use of third-party clinical support for in-house staff, without having to compromise on availability, performance, or accessibility.  

Remote radiology services are increasingly relied upon to support overloaded hospitals, and load balancers are needed to ensure medical images can be easily, securely, and rapidly retrieved from any location, from the hospital's primary and contingency data centers. Dedicated, load-balanced object storage ensures high-performing PACS and other large medical imaging files are immediately accessible to clinicians 24/7, accelerating the delivery of both routine and emergency patient care.

For more information on how to load balance healthcare and object storage applications, check out our detailed deployment guides.