Converged servers are part of a three-tier infrastructure setup that includes storage systems and networking hardware, but hyperconverged server infrastructure does away with this traditional three tier approach. Instead, hyperconverged servers are the base hardware units – referred to as nodes – along with directly attached storage in the form of hard drives and solid state drives built in to the servers.
Additionally, some or all nodes may be “all-flash” nodes to increase storage I/O and reduce latency for those applications which require it.
Also important: Each of these x86 processor-based nodes runs software which allows the hyperconverged servers as well as the compute and network resources to be virtualized to create pools of these three resources.
The writing is on the wall for standalone servers connected to storage resources and dedicated to running a single business application. These servers are being replaced by hyperconverged servers. That is, pools of compute resources that are part of a hyperconverged infrastructure.
To fully understand what hyperconverged servers are, and how they differ from “converged servers” found in converged infrastructure, it’s helpful to explore converged servers and infrastructure.
Converged infrastructure consists of a converged server, along with storage, networking equipment and management software. These are optimized to work together to create a single standardized resource unit which can easily be copied. Vendors of converged servers and other converged infrastructure components include Dell, HPE, IBM, and Fujitsu.
At the simplest level there are two ways to buy converged servers and deploy them as part of a converged infrastructure: using a reference architecture, or by buying pre-configured converged infrastructure units.
- Reference architecture: These are essentially recipes or blueprints for building converged infrastructure units which specify the compute, storage and networking resources to buy, the quantity of these products, and the way to connect them together. The resources themselves may be from different vendors, and the blueprints may have been specifically designed to allow companies to use some of their existing compute or storage resources. The key point of these reference architectures is that they have been pretested and validated, and often include management software designed to simplify further configuration and day to day management.
- Pre-configured units: These units are supplied as data center rack units which can simply be slotted in for very rapid installation and commissioning. Often the units are supplied pre-cabled and connected for maximum convenience, minimum IT staff intervention, very fast deployment and very easy scale-out as demand for resources increases.
Converged Infrastructure and Hyperconverged Server Infrastructure
Hyperconverged infrastructure is often seen as the next step beyond converged infrastructure. But in fact the two are not as similar as the names suggest.
That’s because converged infrastructure generally uses best-of-breed servers, storage, and networking resources to build its infrastructure units. In that respect it provides a way for vendors such as Dell or HPE to package and sell their high-end servers along with other resources in a very easy-to-consume package.
By contrast, hyperconverged infrastructure is more of a software-based solution which includes software defined compute, storage and networking. For that reason it often makes use of low cost commodity servers, or even existing legacy servers, rather than new, high-end branded hardware.
Vendors of hyperconverged servers and other hyperconverged infrastructure include Dell EMC, Nutanix, Cisco Systems, HPE, NetApp, VMware and Huawei.
Virtualized resources refers to the fact that for any particular workload, a complete virtual hardware platform can be configured from hyperconverged servers and their storage with the appropriate compute power, memory and storage resources – including all-flash storage if required.
This can generally be done automatically based on policies for individual applications, without the need for configuring storage constructs such as LUNs. In fact with hyperconverged servers there is no need for any complex storage infrastructure or dedicated storage networks at all.
Scale Up and Out
These hardware platforms can easily be scaled up by adding more virtualized resources from the hyperconverged server pool, and the hyperconverged infrastructure itself can be scaled out by simply adding more commodity servers and storage drives (along with the all important software to turn them into hyperconverged server nodes).
By contrast, converged server environments can be scaled out by adding more resource units but can’t be scaled up without introducing different resource units, which adds to complexity.
Unlike a converged infrastructure, which generally has separate server, storage and network management tools – albeit accessible from a central console – hyperconverged server resources are managed from a single hyperconverged infrastructure administration pane.
Benefits of Hyperconverged Servers
The global hyperconverged infrastructure market size is expected to grow from $7.8 billion in 2020 to $27.1 billion by 2025, (a CAGR of 28.1%) according to a MarketsandMarkets forecast. This strong growth forecast is testament to the many benefits of hyperconverged servers, which include:
- Reduced costs One of the key benefits of hyperconverged servers are lower hardware costs. That’s becasue a hyperconverged infrastructure can use industry standard x86 servers, and does not require high performance storage networks. There is also a risk of vendor lock-in when using servers as part of a converged infrastructure, but this problem is completely bypassed with hyperconverged servers that can be mixed and matched from different low cost suppliers.
- High scalability and greater agility Hyperconverged servers can be bought anywhere and can be set up and put to work in a matter of hours. That makes it much easier for organizations to scale – either up or out – their existing hyperconverged infrastructure than scaling a traditional three tier architecture, or even a converged infrastructure architecture. New virtual servers can also be spun up in a matter of minutes for new applications, allowing organizations to become far more agile. In effect a hyperconverged server infrastructure is a private cloud offering very similar benefits to a public cloud.
- Increased efficiency Like any virtualized server infrastructure, hyperconverged servers increase IT efficiency by increasing server utilization rates. But hyperconverged servers go beyond this by allowing many tedious manual processes to be eliminated, and by allowing a smaller IT team to monitor and manage resources from a single management pane.
- More storage options Hyperconverged servers contain their own storage resources which are pooled and shared with other nodes. That means that the blend of spinning disk and solid state storage can be fine tuned very easily to meet the performance needs of any particular application. This can be achieved without the need for significant capital expenditure on new solid state storage arrays or fast storage networking systems. But it is possible for hyperconverged infrastructure systems to integrate with existing storage systems, so implementing hyperconverged servers does not necessarily make existing SAN (or other storage resources) redundant.
- Disaster recovery Since servers or entire production environments can easily be replicated in hyperconverged infrastructure, these computing environments lend themselves to disaster recovery. In the event of a disaster in one data center, a replica environment can very quickly be brought up in another.