Upgrade or replace? Making informed choices for your IT systems 

Author: Isha Jain

By Tony Hollingsbee, SSD Business Manager, EMEA at Kingston Technology

Across many data centres, investments in IT infrastructure have been held back in recent years due to inflation, increased energy costs and supply chain challenges leading to uncertain delivery times. While this has led some to delay project implementation, others are addressing the need to keep up with customer demands by examining how they can extract greater value from their existing systems.  

The most fundamental issue for data centres is to ensure hardware can manage the increased workloads and volumes of data that are associated with modern technology innovations such as AI, machine learning, IoT, edge and cloud computing. By investing in upgrades for existing systems rather than replacing equipment, data centres are also minimising their carbon footprint. 

Top of the list when considering components that can have an instant impact on existing hardware are storage drives and memory. However, before specifying either, it is vital to first identify the performance and capacity issues that need addressing.  

Boosting memory 

When considering a memory upgrade, the outcome will, at least in part, be determined by existing CPUs, the server platform, and the way memory modules are populated within the server. Memory is always a balance between speed and capacity.  

Often, higher memory bandwidth can be achieved by just populating one DIMM per channel, and adding a second DIMM may result in the memory clocking down. That said, insufficient memory capacity may lead to a greater impact on performance than lower memory bandwidth. 

If data centre servers have free memory sockets, adding in new modules or upgrading from DDR4 to DDR5, for example, could boost performance. Of course, it’s essential to check that the CPU and host system support higher memory bandwidth. If upgrading to DDR5 necessitates a new server platform, the return on investment will be longer, but it will outpace the capabilities of DDR4 in the long term.   

Expanding storage 

If an assessment determines that any bottlenecks in performance come from a lack of storage, rather than memory, then upgrading storage devices would be the right step to take. If the storage set-up is still using HDDs, upgrading to SSDs would result in improved performance. 

Replacing HDDs with SATA SSDs is straightforward to do because they use the same interface and communication protocol (AHCI). SSDs that are built for enterprise and mixed-use workloads will deliver latency and IOPS consistency in data centres and will feature power loss capacitors which will protect data against unexpected power failure.  

Any decisions around upgrading, and future-proofing data centre systems also need to consider the ever-evolving cyber-security landscape. It might be necessary to review security measures as part of the upgrade process, addressing any areas of vulnerability based on past experiences.   

When it comes to making decisions about how to future-proof technology infrastructure, there is a balance to be struck between the financial outlay and the productivity and efficiency advantages of upgrading. Adding any improvement to hardware will boost performance.  

However, upgrading should be considered as a step towards extending the life of servers and maximising ROI, rather than being the ultimate solution. Data centres should be implementing regular maintenance and updating server software and drivers which will also help to increase performance and longevity.  

There are many instances for which upgrading existing infrastructure will be the best option right now to save budget, boost productivity and enable applications including AI and machine learning. With an informed assessment to identify how memory and storage can be best updated, data centres will be able to cater to the needs of users not just today, but into the future.