Reinventing Resilience: How the University of Leicester Transformed Its Data Infrastructure

Every organization, regardless of its size or mission, grapples with the ever-present challenge of managing an exploding data universe. For a large, research-intensive institution like the University of Leicester, home to over 20,000 students and 4,000 dedicated staff members, this challenge isn’t merely about storing files; it’s about safeguarding decades of invaluable research, enabling cutting-edge discoveries, and ensuring the seamless operation of an entire academic ecosystem. It’s a colossal task, and for a while, their existing storage infrastructure was certainly feeling the strain.

Imagine a complex web of aging technology, a data center humming with the tired whir of end-of-life Dell-based Storage Area Network (SAN) backup targets. That was the reality for the University of Leicester. We’re talking about legacy PowerVault and Compellent arrays, a bit like those trusty old cars you’ve had for years – they get the job done, but maintenance costs are climbing, and they’re definitely showing their age. These arrays were intricately connected to ten separate media servers, creating a situation where Logical Unit Numbers (LUNs) were tightly, almost suffocatingly, tied to specific servers. This wasn’t just an inconvenience, you know; it generated significant dependencies. And honestly, it created major headaches and, more critically, posed serious downtime risks. Think about it: if one piece of that complex puzzle faltered, the ripple effect could be devastating, potentially bringing critical university operations to a screeching halt. The IT team, they felt this pressure acutely, especially Systems Specialist Mark Penny, who knew something drastic had to change.

Award-winning storage solutions that deliver enterprise performance at a fraction of the cost.

The Wake-Up Call: A System on the Brink

Traditional SANs, while robust in their prime, often come with inherent limitations for modern, dynamic environments. For Leicester, this meant a rigid, monolithic structure that struggled to keep pace with the university’s burgeoning data needs. The older arrays were nearing, or had already passed, their end-of-life cycles, pushing maintenance costs sky-high and making scalability a genuine nightmare. Adding capacity wasn’t a simple ‘plug and play’; it often involved complex reconfigurations, downtime for migrations, and a continuous struggle with vendor lock-in. Each LUN, acting like a dedicated slice of storage, was meticulously mapped to specific media servers, which is great for predictability in some ways, but it locks you into a very inflexible setup. When you needed to expand or reconfigure, you couldn’t just move things around easily; it was like trying to untangle a giant ball of yarn, each strand crucial and connected to a dozen others. This tight coupling between storage and compute created potential single points of failure everywhere. If a media server went down, or if the underlying LUN encountered an issue, entire backup operations could fail. And in a university setting, failed backups aren’t just an IT glitch; they could mean lost research data, interrupted student services, or even compliance breaches, none of which is acceptable.

Mark Penny and his team recognized the writing on the wall. The existing infrastructure wasn’t just showing its age; it was actively hindering progress and demanding an exorbitant amount of the team’s time just to keep it limping along. They were spending more time firefighting than innovating, and that’s just not sustainable. They needed a solution that was not only robust and scalable but also agile and cost-effective, something that could grow with the university without requiring another rip-and-replace project in a few short years. It was clear: a complete overhaul of their backup platform was essential. They weren’t just looking for a band-aid; they were looking for a fundamental paradigm shift.

A New Horizon: Embracing Object Storage with Cloudian HyperStore

The IT team, driven by a clear vision for a more resilient and efficient future, embarked on a journey to find the right technology. Their extensive research led them to Cloudian’s HyperStore object storage system, a decision that would fundamentally transform their data management capabilities. Why object storage, you might ask? Well, it’s a completely different beast compared to traditional SAN. Object storage treats data as discrete, self-contained objects, each with unique metadata, rather than blocks or files tied to a specific hierarchy. This inherent design offers incredible flexibility, scalability, and resilience, which was exactly what Leicester needed.

One of the primary drivers behind this choice was the burning desire to eliminate the single point of failure that plagued their previous setup. With object storage, data is distributed across multiple nodes, often with built-in redundancy like erasure coding or replication. This means if one node goes offline, the data remains accessible from other nodes, ensuring continuous availability. No more sleepless nights worrying about a single disk array taking down critical systems! Furthermore, object storage naturally streamlines backup processes. Instead of complex LUN management and intricate configurations, you interact with a simple S3-compatible API, making integration with modern backup software far more straightforward and less prone to human error. It truly felt like stepping out of the stone age and into the cloud era, but with the data firmly under their own control on-premises, a crucial requirement for an academic institution handling sensitive research. It wasn’t just about moving data; it was about moving intelligence closer to the data itself, empowering the system to manage its own integrity.

The Path to Transformation: A Detailed Implementation Journey

The deployment of Cloudian’s HyperStore wasn’t an overnight flick of a switch; it was a carefully orchestrated process, a testament to the team’s diligent planning and execution. Let’s delve into the crucial steps that paved the way for this significant infrastructural upgrade:

1. Infrastructure Assessment and Meticulous Planning

Before even thinking about new hardware, Mark Penny’s team undertook a deep dive into their current state. This wasn’t just a cursory glance; it was a comprehensive evaluation that felt a bit like a digital archeological dig. They scrutinized every nook and cranny of their existing infrastructure, meticulously documenting bottlenecks, performance constraints, and, crucially, identifying every single area ripe for improvement. What kind of data was being stored? Everything from confidential administrative records, student information, and academic coursework to terabytes upon terabytes of invaluable research data, potentially spanning decades of scientific inquiry. They had to understand the growth trajectory of each data type, predicting not just immediate needs but anticipating future demands from emerging research fields like AI, machine learning, and high-performance computing, all of which generate astronomical amounts of data. This assessment unequivocally highlighted the pressing need for a far more flexible, massively scalable, and easily manageable storage solution. They also considered the non-negotiables: stringent security protocols, compliance with data protection regulations, and ensuring stellar performance for restore operations, because speedy recovery is just as important as reliable backup, wouldn’t you agree?

2. Strategic Deployment of HyperStore Across Campuses

With a clear roadmap in hand, the Cloudian HyperStore system began its phased rollout. The university has multiple data centers, and the aim was to deploy HyperStore across them, creating a unified, resilient, and scalable storage platform that transcended physical locations. This involved careful network planning, ensuring robust interconnectivity between sites to support the distributed nature of object storage. The team physically installed the Cloudian nodes – those HPE Apollo servers, as we’ll touch on later – configuring them to work in concert. This setup immediately delivered enhanced high availability and built-in data redundancy, directly tackling the glaring vulnerabilities of the previous SAN system. Data isn’t just stored in one place; it’s replicated or erasure-coded across multiple nodes and even different data centers, meaning if a rack goes offline or a data center experiences an issue, the data remains utterly accessible. For an institution where continuous access to research data is paramount, this was a game-changer. The entire process was meticulously documented, ensuring future manageability and troubleshooting would be a breeze, or at least a lot less of a headache.

3. Seamless Integration with Commvault: The Backup Orchestrator

The university wasn’t starting from scratch with its backup software. They already relied on Commvault, a powerful and widely respected data protection platform. The beauty of Cloudian’s HyperStore lies in its native S3 API compatibility, which is the de facto standard for object storage. This made integration with Commvault surprisingly smooth and efficient. It wasn’t about ripping out and replacing; it was about enhancing and empowering. The team configured Commvault to leverage HyperStore as its primary backup target, allowing for seamless data protection and recovery processes. This integration ensured minimal disruption during the transition, as backup jobs could be gradually migrated from the old SAN targets to the new Cloudian system. They focused on phased migration, carefully testing and validating each stage, making sure that every critical dataset was securely backed up and restorable before decommissioning the old systems. This strategic integration brought about a noticeable reduction in complexity and administrative overhead, freeing up the IT team’s valuable time for more strategic initiatives rather than babysitting legacy hardware. It was about making the technology work harder, so the people didn’t have to work harder, if that makes sense.

A Harvest of Benefits: Reaping the Rewards of Modernization

The decision to embrace Cloudian HyperStore wasn’t just about replacing old hardware; it was about unlocking a wealth of operational, financial, and strategic advantages that reverberated throughout the university’s digital landscape. The team watched with satisfaction as the benefits started to materialize, confirming their bold decision was absolutely the right one.

Space Efficiency: A Breath of Fresh Air in the Data Center

One of the most immediate and tangible benefits was the dramatic improvement in space efficiency. In their previous setup, the university needed a whopping 48U of rack space – that’s 48 rack units, each 1.75 inches tall, essentially two full server racks – to house 2.5 petabytes of usable storage. That’s a lot of physical footprint, a lot of power consumption, and a lot of cooling requirements. With HyperStore, they achieved the very same capacity in a staggering 24U. Yes, you read that right: they effectively halved their physical storage footprint. This isn’t just a neat trick; it has profound implications. Fewer racks mean less floor space needed in the data center, which is often a premium commodity. Less hardware means less power drawn from the grid, significantly reducing electricity bills. It also translates to lower cooling costs, as there’s less heat generated. This kind of efficiency isn’t just good for the budget; it’s also a win for the environment, aligning with broader sustainability goals that many institutions are now prioritizing. Imagine the sigh of relief from the facilities team when they saw the consolidated footprint; it probably felt like they’d gained a whole new room!

Cost Savings: More Than Just the Hardware Price Tag

The university isn’t just saving on floor space and power; they wisely anticipated a substantial 25% reduction in overall data storage costs upon full implementation of the Cloudian-based solution across all three of its data centers. This isn’t just about the initial purchase price, either. The cost drivers in traditional SAN environments are notoriously complex: think expensive annual maintenance contracts for end-of-life hardware, escalating software licensing fees that often scale with capacity, the aforementioned power and cooling expenses, and the constant need for disruptive, costly upgrades every few years. Object storage inherently tackles many of these issues. Its architecture allows for linear scalability – you add nodes as you need them, without having to overprovision for future growth. The licensing models are often more flexible, and the commodity hardware approach significantly reduces capital expenditure compared to proprietary SAN appliances. Over time, the Total Cost of Ownership (TCO) shifts dramatically in favor of object storage, freeing up valuable budget for other strategic IT initiatives, perhaps even allowing for investment in new research computing capabilities.

Enhanced Data Availability: Peace of Mind for Critical Research

For a university, data isn’t just files; it’s the lifeblood of research, the foundation of knowledge, and the repository of academic achievement. The HyperStore system provided robust, built-in redundancy, finally eliminating those terrifying single points of failure that had loomed large. This meant continuous access to backup data and, crucially, minimized potential downtime. Cloudian’s architecture utilizes either N-way replication (where every piece of data is copied multiple times across different nodes) or erasure coding (a more space-efficient method that breaks data into fragments and distributes them with parity information, allowing reconstruction even if some fragments are lost). This sophisticated approach ensures that even if several nodes fail, the data remains intact and accessible. For researchers working on grant-funded projects, knowing their critical datasets are perpetually available and protected is invaluable. Imagine the stress if a thesis project’s data vanished or a multi-year research initiative ground to a halt because of a storage outage. Now, that anxiety has been largely replaced with confidence, and truly, that’s priceless.

Simplified Backup Processes: Regaining Control and Time

The integration of HyperStore with Commvault didn’t just make things more efficient; it fundamentally streamlined backup operations, drastically reducing complexity and administrative overhead. Previously, the team often found themselves mired in the minutiae of managing those individual LUNs, constantly provisioning and re-provisioning storage, troubleshooting connectivity issues, and manually managing backup jobs that felt like they were held together with string and duct tape. The old system was prone to errors, and just keeping track of everything was a full-time job. With the new object storage system, much of this burden evaporated. The system’s self-contained, policy-driven nature made it significantly easier to manage and, perhaps even more importantly, much simpler for staff to learn and operate, even those who might have been initially unfamiliar with such cutting-edge solutions. Imagine the joy of setting a policy once and knowing your backups are happening reliably, without constant hand-holding. This shift allowed the IT team to move away from reactive troubleshooting to proactive management, fostering a more strategic approach to data infrastructure.

Performance Gains: Speeding Up the Essential

While the initial article didn’t explicitly detail performance improvements, it’s a natural and expected outcome when migrating from aging SANs to a modern, distributed object storage system. The legacy Dell arrays likely suffered from bottlenecks in I/O performance, especially when handling large volumes of concurrent backup and restore operations. Object storage, particularly when running on commodity hardware configured for scale-out performance, can dramatically improve data throughput. This means faster backup windows, allowing backups to complete within their allocated timeframes, reducing the risk of conflicts with production systems. Crucially, it also means significantly faster data recovery. In an incident, being able to restore terabytes of data quickly can be the difference between a minor disruption and a major crisis. The distributed nature of Cloudian across multiple nodes allows for parallel processing of data, greatly enhancing both read and write speeds for large datasets, which is absolutely vital for a busy academic environment constantly generating and consuming information.

The Unsung Hero: High Availability with Load Balancer.org

Achieving true enterprise-grade high availability and seamless scalability isn’t just about selecting a great storage platform; it’s also about intelligently directing traffic to it. The University of Leicester didn’t stop at Cloudian; they smartly implemented Loadbalancer.org’s HyperBalance appliances to supercharge the system’s reliability and performance. These appliances weren’t just an afterthought; they were a critical component, acting as the intelligent traffic cops for their data flow.

They deployed a high-availability pair of these HyperBalance appliances, strategically placed across two data centers. This configuration is key: if one load balancer unit were to fail, the other would seamlessly take over, ensuring zero interruption to services. These powerful devices were tasked with balancing the incoming traffic across no less than 15 HPE Apollo servers, which serve as the actual Cloudian storage nodes. Think of it like a perfectly choreographed dance: the load balancers efficiently distribute incoming requests and data transfers, ensuring no single server gets overwhelmed and that every component is utilized optimally. This setup was managing an average of 120 terabytes of data per week, a staggering volume that speaks volumes about the university’s data generation. The HyperBalance appliances were ensuring the continuous high availability of the Cloudian HyperStore solution, a mission-critical role.

Mark Penny, clearly impressed, once noted, ‘We have been very impressed with the way that the Loadbalancer.org solutions have reacted to storage node failures during testing. The failover occurs so quickly that storage jobs don’t even know that an incident has occurred.’ Now, that’s a statement that would bring a smile to any IT professional’s face! What does ‘storage jobs don’t even know’ really mean? It means the load balancer detected the node failure, instantly redirected traffic away from the faulty server to the healthy ones, and did it all in the blink of an eye. The backup or restore jobs running on Commvault continued uninterrupted, completely unaware that a piece of the underlying infrastructure had momentarily stumbled. This level of seamless failover is the holy grail of high availability, preventing errors, retries, and, most importantly, avoiding any disruption to critical data operations. It’s a testament to robust engineering working in harmony.

Looking Ahead: A Foundation for Future Innovation

The University of Leicester’s successful implementation of Cloudian’s HyperStore and Loadbalancer.org’s HyperBalance appliances has done more than just fix an aging problem; it has laid a robust and incredibly agile foundation for future growth. The scalable nature of this new system positions the university to comfortably accommodate the inevitable increase in data volumes without significant, disruptive additional investment. As research continues to push boundaries, particularly in fields like genomics, climate modeling, and artificial intelligence, the demand for storage will only intensify, and this infrastructure is built to scale out, not up. You simply add more nodes as needed, which is a much more flexible and cost-effective approach than the forklift upgrades of yesteryear.

Penny’s concluding observation perfectly encapsulates the team’s satisfaction and confidence: ‘We did a lot to try and break it. It wasn’t a problem at all. Everything just worked.’ This isn’t just a casual remark; it speaks volumes about the rigorous testing they put the new system through, pushing its limits to ensure its resilience under stress. When a complex, mission-critical system performs flawlessly under such scrutiny, it instills a deep sense of trust and reliability within the IT team and, by extension, across the entire university community. It’s that feeling of quiet confidence when you know your tools are truly up to the job, no matter what challenges may come.

This case study beautifully underscores the paramount importance of selecting flexible, scalable, and resilient storage solutions to meet the evolving and often unpredictable needs of modern educational institutions. By wholeheartedly embracing innovative technologies and meticulously planning their deployment, the University of Leicester hasn’t just upgraded its data management capabilities. They’ve genuinely transformed them, ensuring the protection, accessibility, and integrity of critical research data, administrative records, and learning resources for many, many years to come. It’s a prime example of how strategic IT investment can not only solve present problems but also proactively empower an institution’s future, ensuring that the pursuit of knowledge continues unimpeded, which, at the end of the day, is what a university is all about.

References

• Cloudian. (2019). Case Study: University of Leicester Uses HyperBalance. Retrieved from cloudian.com
• Cloudian. (2019). Case Study: University of Leicester. Retrieved from cloudian.com
• Computer Weekly. (2019). University of Leicester dumps SANs and LUNs for Cloudian object storage. Retrieved from computerweekly.com
• Inside HPC & AI News. (2019). University of Leicester Adopts Cloudian Object Storage for Backup. Retrieved from insidehpc.com
• Loadbalancer.org. (n.d.). Case Study: University of Leicester. Retrieved from pdfs.loadbalancer.org