There are a lot of moving parts to modern datacenter infrastructure. Beyond the metal boxes that make it all go, there are software components and services that have become so fundamental as to be part of the baseline consideration for any new deployment. Unless you have all the pieces you cannot assemble a datacenter efficient enough to be competitive.
The march of progress has moved us far beyond mere FLOPS per watt (power efficiency) or watts per square metre (power density). In today’s world the measure of efficiency is more ephemeral, with Operational Expenses (OpEx) mattering more at scale than Capital Expenses (CapEx).
With very few exceptions, workloads are no longer CPU bound. DRAM density remains a standard candle. Storage efficiency – measured in effective storage per cluster as well as IOPS per cluster – is increasingly a critical measurement. The speed, efficiency and reliability of backups and data protection mechanisms are critical, as is the ease of use of the management tools that bind it all together.
In today’s software-defined, everything-as-a-service world, datacenter resources are supplied to customers in a utility model. That means the infrastructure running your workloads needs to be as reliable as electrical provisioning through the national power grid.
It is the interplay between hardware management, data management and software management that determine the OpEx of a datacenter. Well tested, well integrated solutions that “just work” are absolutely essential to realising a profit. Downtime is expensive and there is no room for unexpected complexity.
Simplicity matters. Your infrastructure must be more than a series of disparate components lashed together under a single marketing banner and pushed out as a stack to gain a vendor those extra points of margin. Your infrastructure needs to be modular, resilient, interchangeable and known good.
This, at its core, is the argument for convergence, and over the years convergence has evolved.
In the beginning there was the Vblock, and the Vblock was good. Vblock (and eventually Flexpod) reduced the complexity of negotiating with multiple vendors for the equipment everyone knew was going to end up purchased anyways. Servers, storage, an SSD acceleration tier, switching and a hypervisor were bundled up into a nice neat package and sold as one.
The stack was tested as a unit and sold as a single block with unified management. These stacks were the best of the best, designed by the best and sold at a premium price. This was the first generation of modern converged infrastructure, or “convergence 1.0”.
Convergence 1.0 stacks had the benefit of combining equipment from enterprise vendors with pedigree. These vendors were expensive, yes, but you knew who the vendors were, how their software worked, and all the tics and foibles that went along with them. They were the safe bet.
Not long after Vblock and Flexpod made their splash a new generation of vendors came onto the scene. Servers, storage, SSD acceleration, storage switch and hypervisor were all collapsed into a commodity hardware cluster and handled in software. This offered both lower CapEx and lower OpEx when compared to Convergence 1.0 stacks.
Convergence 2.0 stacks had only one vendor to deal with, one management console to learn, and the hardware was cheap, disposable and interchangeable. Instead of swapping out whole racks full of workload-specific gear, individual nodes became as simple to use as Lego.
Need more capacity? Add some nodes and you would grow your storage and your compute at the same time. Dead node? Pull the whole thing and replace with virtually zero fuss and muss. Convergence 2.0 married storage and compute and made it easier to manage than replacing disks in a RAID array.
Convergence 2.0 wasn’t just cheaper, it was simpler…and simplicity is good.
Despite the admirable and impressive achievements of the convergence 2.0 vendors, critical infrastructure components were missing from the stack. Servers, storage, SSD acceleration, storage switch and hypervisor are only part of the story.
Convergence 2.0 vendors will let you light up a VM and move it from neighbour to neighbour simply, easily and in a cost effective fashion. What are missing are the layers of data management and protection that have become so fundamental to today’s datacenters.
SimpliVity has realised the evolution of convergence, building upon the work of its predecessors by incorporating data management and protection. SimpliVity’s converged stack includes a backup dedupe appliance, WAN optimisation, a cloud gateway, a storage caching appliance, data protection apps and global management across all layers, including automation.
SimpliVity offers lower costs, not only through the inclusion of these critical technologies into the stack, but through data efficiency obtained via deduplication, compression and write optimisation. SimpliVity incorporates multiple tiers of storage: DRAM, Flash, HDD and Cloud, making sure you get the IOPS you need when you need them and where you need them.
From snapshots and cloning to backups and disaster recovery, the SimpliVity stack offers the fastest and simplest global data management solution available today. SimpliVity is today’s critical datacenter infrastructure, presented as a single integrated stack, designed and tested as one and deployed as a whole.
This is convergence 3.0.
Solving tomorrow’s problems today
Data and workloads in the datacenter need to do more than flow from one node to another. They need to do so safely. This must transition seamlessly from one side of the world to the other and back again without need to fret about the complexity of backups, disaster recovery, capacity or speed.
With SimpliVity there is no requirement to learn a completely new UI. Everything is a single point of management: VMware’s vCenter. Complexity kills. Book a demo today and find salvation in the simplicity of SimpliVity and convergence 3.0.