Enterprise Storage Forum content and product recommendations are editorially independent. We may make money when you click on links to our partners. Learn More.
Servers host enterprise workloads and provide storage, networking, and compute resources to data centers.
They have power that ordinary computers don’t, allowing incredibly heavy workloads to run on enterprise machines and networks.
As the data management market grows, servers continue to undergo development to meet ongoing demand for increased data center capacity and speed.
The following server trends include form factors, central processing unit (CPU) advancements, and methods to improve data center efficiency:
6 Trends in the Server Market
- New Server Processors
- More Server Consolidation
- Rapid Growth of the Global Microserver Market
- Expanding Software-Defined Server Solutions
- Driving Two-Phase Immersion Cooling in Servers
- Looking to Server Modularity
1. New Server Processors
Enterprises are focusing on computing improvements to server hardware, including processors.
Processors are the nucleus of a computing machine, holding together all hardware interactions, so server improvements and market growth often come from improvements made to processors.
In spring 2021, AMD announced a new server processor family: EPYC 7003. For single-core applications, EPYC 7003 doubles memory access capabilities. AMD’s new processor family is useful for high-performance computing (HPC) applications as well as containerized workloads, according to Forbes.
Intel and AMD are also amping up their server processors for 2022. These updates are projected to include PCIe 5.0 interfaces, according to the Dell’Oro Group. In October 2021, Intel demonstrated high-bandwidth memory (HBM), which will feature in its fourth-generation Xeon processors.
NVIDIA’s EGX platform, first announced two years ago, is a leader and is offering artificial intelligence (AI) for the edge with a graphical processing unit (GPU)-centric architecture, rather than CPU-focused. AI workloads run on NVIDIA partner servers with NVIDIA GPUs. GPUs are the future of advanced AI, not CPUs: the modern CPU just won’t be able to do enough, according to Zeus Kerravala, acclaimed IT analyst and researcher.
“The demands of AI go far beyond what CPUs are capable of achieving,” Kerravala says.
NVIDIA’s EGX solution doesn’t just provide a GPU: the platform is a software-defined infrastructure that combines compute with networking and hardware with software.
“There are many makers of GPUs today, but what keeps NVIDIA out in front is its ability to deliver a ‘full stack’ to simplify the process of deployment.”
2. More Server Consolidation
Many server form factors are designed with high power and data center density in mind.
The more servers that can be crammed into a physical space, the more power that can support high-performance applications and workloads.
This trend is continuing.
“Hardware companies explored consolidating numerous physical servers in a smaller, more manageable area before virtualization became popular,” said Axel Hernborg, CEO of Tripplo, a Swedish travel planning site. “The HP c-Class blade server chassis, which combined servers and storage in a blade style, shook up the server market five years ago. Since then, no other device has had the same commercial effect.”
Now, not only blade servers and rack servers but also microservers — chip-style servers that slot into a chassis — conserve space, so data centers can provide more power.
3. Rapid Growth of the Global Microserver Market
Microservers have been used for running websites, but they’re taking off in other arenas too.
Recent research on microservers in hyperscale computing environments indicates their suitability because of their low power consumption and performance: hyperscale environments benefit from maximized energy and high-performance computing, according to MarketsandMarkets.
The firm expects the global microserver integrated circuit (IC) market to have a compound annual growth rate (CAGR) of 12.6% from 2021 to 2026.
The fastest microserver processors are ARM-based. Though microserver applications have been previously limited, the report suggests that developments like 64-bit ARM processors make microservers a better candidate for important workloads.
4. Expanding Software-Defined Server Solutions
A software-defined server manages multiple servers’ resources from one location, abstracting them so that they’re more widely available. Those resources include software-defined storage systems. Prepackaged software-defined servers allow enterprises to run workloads and storage from multiple places.
Dell, for instance, is prepackaging servers with compute, which can then be shipped to remote or edge locations. These solutions run software-defined storage (SDS) through scalable servers and storage nodes. Software-defined data center infrastructure manages resources, including servers, more efficiently than manual or hardware-based architectures.
5. Driving Two-Phase Immersion Cooling of Servers
Some data centers, Microsoft’s among them, are using two-phase immersion cooling, a form of liquid cooling, as a key approach to thermal management.
To cool processors, electronic-safe fluid, not water, boils at a lower temperature than water, and the resulting vapor hits a cooled condenser, turns back into liquid, and falls on the servers.
For Microsoft, this is preparation for a time when air cooling won’t be sufficient. It means that, if successful, computing resources could be placed closer together again, rather than distant to maintain sufficient airflow. Two-phase immersion cooling also helps reduce energy consumption.
6. Looking to Server Modularity
Modular infrastructures and systems allocate compute, networking, and storage resources independently for more efficient resource use.
Modular blade servers, by vendors such as Intel and Mercury Systems, are designed for precision and more customizable for edge application needs.
A modular chassis — as in Dell PowerEdge Modular Infrastructure — has scalable modules for compute, networking, and storage resources. This scalability allows enterprises to use only the resources their existing workloads need, saving them money.
Modular platforms also have the potential to connect the latest technologies to data center resources.
“Modular systems will offer high-speed interconnects that provide the ability to shape resources to the needs of workloads and seamlessly integrate new technologies, such as next-generation processors, accelerators, persistent memory, and 400 G networking,” said Tim Stack, senior marketing manager at Cisco.
Read Next: Best Server Security Tools
