IT Solutions - Sep 29, 2022

New technologies require flexible servers.

“We need three more graphics cards right now!” Sound like a familiar demand from your developers? New technologies such as GPU-intensive artificial intelligence require increasingly flexible infrastructures as they can quickly push traditional rack and blade servers past their limits. In the UCS X platform, Cisco is now combining the best of both architectures.

Written by

Team Leader Specialist Sales

E-Mail: gerald.rabeneck@bechtle.com

Many IT teams are all too familiar with the problem. While time is racing away and development needs are becoming increasingly difficult to predict, the adaptability of server room infrastructure remains fairly limited. Rack servers are commonplace as they provide plenty of capacity for CPUs, GPUs and hard drives, but they are quite inflexible. While it is technically possible to scale up a rack, this isn’t really practical because it requires complex modifications and means extended downtimes. If additional IT resources are what it takes, the answer tends to be to buy a new server, which costs both time and money and makes it impossible to react quickly to your developers’ needs.

IT teams requiring greater flexibility therefore leverage blade servers. These come with swappable enclosures that share the power supply and data lines within the chassis, facilitating the integration of new modules. However, blade servers do have one disadvantage and that is they only have limited capacities for hardware such as RAM modules, graphics cards and CPUs. They may be a flexible solution, but they will also quickly reach the maximum of their capabilities.

Many IT teams are all too familiar with the problem. While time is racing away and development needs are becoming increasingly difficult to predict, the adaptability of server room infrastructure remains fairly limited. Rack servers are commonplace as they provide plenty of capacity for CPUs, GPUs and hard drives, but they are quite inflexible. While it is technically possible to scale up a rack, this isn’t really practical because it requires complex modifications and means extended downtimes. If additional IT resources are what it takes, the answer tends to be to buy a new server, which costs both time and money and makes it impossible to react quickly to your developers’ needs.

IT teams requiring greater flexibility therefore leverage blade servers. These come with swappable enclosures that share the power supply and data lines within the chassis, facilitating the integration of new modules. However, blade servers do have one disadvantage and that is they only have limited capacities for hardware such as RAM modules, graphics cards and CPUs. They may be a flexible solution, but they will also quickly reach the maximum of their capabilities.

A cellar full of specialists.

Unfortunately, the shortcomings of a traditional platform aren’t only felt on a technical level, but also in day-to-day strategic operations. The reason? Servers are typically purchased to perform specific tasks which makes it difficult for CIOs to create resource pools. So, for example, a server that has been used for the Oracle database cannot simply be used for a Virtual Desktop Infrastructure because it doesn’t have the required GPU power. This sort of specialised server capabilities means that data centres often have available capacity, but not usually on the servers where it is needed right now.

Cisco UCS X – Modularity meets capacity.

In the UCS X (Unified Computing System X) server platform, Cisco has developed a sophisticated solution to meet this challenge head on. The system combines the best of rack and blade servers such as large-capacity, swappable modules and shared resources. One aspect that really stands out is the architecture’s flexibility. Cisco has completely done away with the mid and backplanes meaning that IT admins can add CPUs, GPUs and storage in the chassis as desired and can also choose if they want to use air or water cooling. Connecting the individual components is also simple using Intelligent Fabric Modules (IFM) and Virtual Interconnect Cards (VIC) via a 400G/1600G fabric which does away with the need to pre-define connections for each slot.

What’s more, Cisco has arranged the nodes vertically instead of horizontally to create more space for compute, storage and networking power. The swappable modules make it easy to quickly scale the UCS X to business needs and the integration of a new graphics card, for instance, is a matter of mere minutes. In other words, the UCS X architecture is opening the door to greater freedom and future-proof IT landscapes as the system grows flexibly with the needs of your IT department.

Sustainable ventilation concept.

It’s not just system flexibility and performance that are important for CIOs and IT admins, but also energy consumption as energy costs rise and Green IT becomes an integral part of corporate sustainability strategies. UCS X’s efficient ventilation system can support energy saving, but to understand how they do it, it’s important to know that common server architecture fans all work at the same speed.

If one is rotating at high speed to cool a particularly hot slot, the others will do, too, meaning the entire chassis is cooled right down. Cisco’s engineers have designed individually controllable fans and built them into the UCS X to save a significant amount of energy. Should a CPU need to be cooled down a bit more, the nearest fan will increase its rpm while the others carry on as normal.

Intersight – Much more than management.

UCS X servers are managed through Cisco Intersight, which provides IT admins with a single pane of glass for managing and monitoring hybrid cloud resources and can be easily integrated into existing infrastructures. But that’s not all. It also offers integrated tools IT teams can leverage, for example, to optimise VM performance and simplify complex workflows either via drag-and-drop or automatically. In short, UCS X and Intersight deliver desperately needed flexibility to server rooms, allow IT admins and CIOs to respond better to developer needs and ready themselves for the technology of the future.

Share this page

This post was published on Sep 29, 2022.