We look at five hyperconverged systems for growing businesses seeking to simplify their IT hardware needs.
For many small businesses, one file server or network attached storage (NAS) appliance might be all you need, or you could even keep all your files centralised online with the likes of Dropbox Business. However, for fast-growing or mid-sized businesses, your IT hardware needs may be more complex and becoming increasingly difficult to manage.
Hyperconvergence provides a way of simplifying and scaling IT infrastructure with pre-integrated building blocks from a single supplier. It’s not the only solution – and we’ll explore another option, a cloud infrastructure-as-a-service (IaaS) provider such as Amazon Web Services, next month.
There are good reasons to keep your systems out of the cloud, however, and hyperconverged infrastructure is becoming increasingly popular alternative. Very small offices might be better off with an entry-level server – which we’ve covered in another feature – but lower-end hyperconverged products can be a feasible solution for fast-growing or mid-sized businesses looking to upgrade or simplify their setup.
But how exactly does hyperconvergence differ from traditional hardware solutions? And what are the options? Let’s start with the first question.
What is hyperconvergence?
To understand hyperconvergence, we need to go back and look at its predecessor, converged infrastructure. Rather than buying servers, storage arrays and network switches and then assembling them into a system yourself, the idea was to buy a pre-integrated rack of equipment incorporating all those elements, plug it into power and network connections, and off you’d go – at least in theory. A well-known example was the VCE Vblock.
The concept could be taken a step further by the vendor optimising the hardware for particular purposes and installing the relevant software at the factory, as Oracle does with its engineered systems family.
Hyperconverged infrastructure differs from converged by using all-purpose ‘commoditised’ servers and doing away with the storage array altogether. Storage is still required, but it is attached to the servers, and a software layer makes the data stored on each one available to the wider system.
How it works
As such, hyperconvergence is seen as part of the move towards ‘software-defined’ technologies – where software assumes a more important role in controlling IT infrastructure and the hardware itself is commoditised.
The hyperconverged software layer works alongside virtualisation – the well-established technology used to create ‘virtual machines’ that are separated from actual hardware. This gives you the flexibility to pool your hardware resources and allocate them more precisely for specific tasks and applications.
Hyperconvergence needs more processing power than would otherwise be the case, but the cost benefits of using commoditised hardware more than make up for that. It’s also modular and relatively easy to scale, so you can just add one or more hyperconverged systems, or ‘nodes’, as your needs grow.
Hyperconverged products are available from recent entrants specialising in this type of system, such as Nutanix and Pivot3, as well as traditional vendors such as Dell EMC (which also sells systems based on Nutanix software), Hitache Data Systems, NetApp and HPE (which recently acquired SimpliVity).
Next: Some of the more affordable hyperconverged systems compared.
Dell EMC's VxRail family starts with the G series. It runs VMware's virtualisation software on Dell EMC hardware with nodes featuring single or dual Xeon E5-2600 processors, at least 64GB of RAM, six drive bays (with a minimum of 200GB of cache SSD, 3.6TB of hybrid storage or 3.84TB of all-flash storage), and either two 10Gb Ethernet or four Gigabit Ethernet ports. Between three and 64 nodes can be clustered.
Dell EMC also sells systems using the Nutanix Xpress software running on PowerEdge servers. The XC430 Xpress is a one-node system with dual Xeon processors, at least 64GB of RAM, one SSD (400GB to 1.9TB), up to three hard drives (2, 4 or 6TB each), and at least one Gigabit Ethernet port. Three or four nodes are deployed together. The XC430 Xpress is said to be suitable for virtual desktop infrastructure (VDI), SQL Server and other workloads.
Hitachi Data Systems (HDS)
The HDS Unified Compute Platform (UCP) HC (hyperconverged) V240 has a minimum configuration of one eight-core Xeon processor, 32GB of RAM, two 10 Gigabit Ethernet ports, 3.6TB of hybrid storage (all-flash models in the UCP HC family provide deduplication and compression) with a 400GB SSD for caching, and VMware virtualisation software.
A minimum of two nodes is required, and up to 64 nodes can be clustered. HDS is somewhat vague about the capacity of the V240, saying only that “the UCP HC appliance is sized to run approximately 120 average-sized, general-purpose, data centre VMs or 250 virtual desktops with no restrictions on application type” without qualifying that statement in terms of the model (V240, V240F, V210, V210F) or configuration.
Earlier this year, HPE acquired hyperconvergence specialist SimpliVity. The HPE SimpliVity 380 is a hyperconverged node based on the HPE ProLiant DL380 server. Specifications include dual Xeon E5 processors (allowing eight to 22 cores per node), dual 10GbE ports, and a minimum of 256GB RAM and five 1.92TB SSDs. Those SSDs give an effective capacity of 6-12TB, depending on the data being stored, thanks to compression and deduplication. Each node supports a "large number" of VMware virtual machines.
A minimum of two nodes is recommended, and up to 32 nodes can be used together. The number of nodes in a cluster can be adjusted up or down as required.
A “small” NetApp hyperconverged system is somewhat larger than most of the other products mentioned here, comprising at least two compute nodes and four storage nodes.
A small compute node has 16 cores, 256GB of RAM, four 25/10GbE and two Gigabit Ethernet ports, while a small storage node contains six 480GB SSDs with an effective capacity of 5.5 to 11TB. Small, medium and large nodes can be mixed and matched as required.
Nutanix Xpress is specifically aimed at small and mid-sized businesses. Each appliance can be configured with three or four nodes. Each node has 16 processor cores, at least 64GB of RAM, 480GB (or larger) SSD, dual hard drives (at least 2TB each), and Gigabit Ethernet. A minimal system is said to be good for running up to 10 virtual machines. At the other extreme, a fully spec'ed Xpress can run 100 virtual machines.
Nutanix's software includes support for the VMware vSphere and Microsoft Hyper-V virtualisation platforms, as well as simplified management with sophisticated analytics, and multiple data protection capabilities. The storage subsystem includes deduplication, compression and automatic tiering.