HCI Appliance – Time For a Change
Hyper-Converged Infrastructure – HCI for short – is today a well-established technique in enterprise computing. In the last 5-10 years, the software for HCI has matured rather nicely, from the basic hypervisor layer all the way through extensive management and orchestration layers. Today, that includes several forms of managing workloads in a hybrid environment, specifically sharing and moving workloads – virtual machines and their data – between on-premises HCI kit and the public cloud.
However, one facet of HCI has not evolved so rapidly – namely, the actual platforms used to execute the software. By any other name, we call these servers – and they have only incrementally improved over the last decade. We now have well-known server suppliers into double-digits in terms of ‘generations’ of product. Consider this – when was the last time you used, say, the 14th generation of anything?
Today, it’s time to reposition HCI into a powerful force within enterprise computing, in terms of the actual workloads it can process efficiently. Sadly, enterprises have tended to implement HCI software on rather bland platforms, with (again sadly) all-too-predictable, mediocre results. Datacenters now contain dozens, hundreds, or even thousands of garden-variety servers, each running an instance of HCI software – at a high cost, since the stacks are typically licensed per socket. The proliferation of sockets means a corresponding rise in software licensing costs, in particular, costs for software-defined storage capability, something inherent to HCI.
To overcome that dilemma, the answer is to move to much more powerful sockets and platforms than before – in other words, as some might say, “up your HCI game” by implementing the now-substantial HCI software capability on an equally substantial platform, such as Axellio’s FabricXpress™ HCI Appliance.
The New Player in HCI
The FabricXpress HCI Appliance is ideal for running 2-3x the sheer number of virtual machines as other standard server platforms, just on a socket (processor) capability. Unlike garden-variety servers, Axellio’s HCI Appliance uses only “top-bin” CPUs, which provide a much greater ‘bang for the buck’ than mid- or low-bin CPUs. After all, if you buy a CPU, keep it busy! The cores inside that CPU care not if you execute idle loops or useful work; you are providing power and cooling for it all the same.
Beyond sheer compute horsepower, however, lies the biggest issue with standard servers running HCI stacks today – namely, data for compute, and the necessary storage. This is where the FabricXpress HCI Appliance really shines, in its ability to hold just shy of a petabyte – yes, 920 TB – of the fastest, densest block storage devices available today, namely SSDs with the NVM Express transport protocol – and all in just the starting configuration taking up only 4U of rack space. All commonly used HCI stacks support such SSDs, but the FabricXpress HCI Appliance supports more of them, with independent of compute scaling, at native speeds, with greater reliability, and higher efficiency and (most importantly) superior TCO than commodity servers– exactly what software-defined storage HCI capability needs.
Leveraging an Established Player in HCI Software
To that end, consider the Microsoft ‘Storage Spaces Direct’ stack, as implemented inside Windows Server (Datacenter Edition) 2016 and 2019. Storage Spaces Direct is the ideal stack to run on the FabricXpress HCI Appliance for HCI software-defined storage.
It can leverage native RDMA capability via SMBv3 for the highest performance possible, across a cluster of nodes, as well as providing outstanding management and orchestration capability via the native Windows Server Manager (GUI) or, better yet, Powershell commands.
Best of all, however, is the fact that Storage Spaces Direct is built-in, requiring no extra licensing fees, unlike some ‘virtual storage area network’ capabilities found in other HCI stacks. Effectively, using Storage Spaces Direct, one can harness all the power inherent in the FabricXpress HCI Appliance for the most efficient HCI user experience possible. Imagine running 100’s of virtual machines per 4U instead of 25, and running all those 100+ VMs significantly faster, with world-class block storage under control of a highly scalable, coordinated, orchestrated data and storage management framework, built-in to the software stack. All at a TCO that is demonstrably superior to other server platforms with other HCI stacks.
It’s Time to Up Your HCI Game
Now that we’re here in 2018 and HCI software is very robust and well understood, it’s time to up your HCI game with a better HCI appliance. Incremental improvement in standard servers is nice, but significant, factors instead of incremental improvement are rare. The industry just doesn’t usually “work” that way. It takes invention and discovery to break through. Remember, electricity did not come from the continuous, incremental improvement of candles.
AXELLIO has broken the paradigm of incremental improvement with the FabricXpress HCI Appliance, and in combination with Windows Storage Spaces Direct, is the new standard by which all HCI implementations will be measured. So, where do we go from here? We go to the next level of HCI, with FX-WSSD, paying homage to the past incremental improvements while at the same time, leaping over them. The HCI future is now. Run workloads you could only dream of before, at efficiencies previously not possible. Up your game!