Hyperconverged infrastructure has quickly become one of the hottest trends in IT today. There are dozens of companies, from startups to established vendors, touting systems that integrate compute, storage, virtualization resources, networking and other software components.
Howard Marks, Network Computing contributor and Interop speaker, noted that the term "hyperconverged" was coined in 2012 by former VMware CTO Steve Chambers. Storage consultants the Taneja Group also lay claim to the name, supposedly to describe Scale Computing's offering. Whatever the origin, it's used to differentiate new integrated appliances from startups like Nutanix from converged systems like VCE's Vblock. The introduction of VMware's EVO:RAIL platform in August brought hyperconverged systems into the mainstream lexicon.
Hyperconverged systems now make up a fast-growing part of the overall converged infrastructure market, which was estimated by Gartner (which refers to it as integrated systems) to exceed $6 billion in revenue last year.
The first generation of converged infrastructure products integrate storage arrays with servers and networking, Wikibon Principal Research Contributor Stuart Miniman wrote in a January blog post. "While these solutions offer simplicity and better economics than traditional infrastructure, there has been an opportunity to take advantage of the latest technology (including flash storage, services in software and increased density of cores) to create a new architecture that both reduces the components of infrastructure while enabling distributed solutions."
Learn more about the changing storage landscape in the Storage Track at Interop Las Vegas this spring. Don't miss out! Register now for Interop, May 2-6, and receive $200 off.
So what makes a hyperconverged system different from a converged system? In an InformationWeek blog post, Bhargav Shukla, director of product research and innovation at KEMP Technologies, described converged systems as typically consisting of "separate components that are designed to work together" compared to hyperconverged systems, which he said "are truly modular, resulting in a scale-out approach as opposed to a scale-up one."
These hyperconverged systems are usually delivered in an appliance form factor, he added.
According to SimpliVity, an early hyperconvergence player, there are a few defining features that distinguish truly hyperconverged systems, including a focus on the virtual machine, the ability to easily scale out resources by adding x86 nodes, native data protection, and a software-centric design. These systems are developed and supported by a single vendor, per SimpliVity's definition.
The ultimate goal of hyperconvergence is to enable IT departments to improve efficiency, reduce operational complexity, and quickly scale to meet growing requirements.
Lots of vendors are jumping on the hyperconvergence bandwagon. As part of our inaugural "Top in Tech" series, we've picked 10 that are leading the way and making their mark in this new hyperconverged world.
Hyperconvergence pioneer Nutanix launched its first product in 2011 and initially focused on a message of "ban the SAN." Today, the company's Virtual Computing Platform provides integrated compute and storage through servers running a standard hypervisor and the Nutanix Operating System. According to Gartner's report on integrated systems, Nutanix's technology is unique in that "the storage and compute elements are natively converged to create a much tighter level of integration" -- a node-based approach that "enables theoretically limitless additions of new compute or storage bandwidth in very small increments."
Nutanix, which released what it claims was the industry's first all-flash hyperconverged array last year, has raised $317 million in funding, filed 43 patents, and touts an annualized sales run rate of $300 million. Last year, the company inked an OEM deal with Dell to offer converged appliances built with Nutanix software running on Dell PowerEdge servers.
Another pioneer in the hyperconvergence space, SimpliVity came out of stealth mode in 2012. The startup's OmniCube platform combines compute, hypervisor, storage services and network switching on x86 server hardware with centralized management. OmniCube goes further than other integrated systems by incorporating features such as built-in VM backup, in-line data deduplication, compression and optimization at the source, according to Gartner.
Scale Computing @ScaleComputing
Scale Computing differentiates its HC3 and HC3x hyperconverged platforms with ease of use and simplicity, according to Scott D. Lowe writing at Wikibon. The company chose to standardize on KVM as a single hypervisor and built its own management layer, creating high functionality without the need for a virtual storage appliance, coupled with object-based storage with direct access to the hypervisor. Reliance on KVM also eliminates licensing fees for commercial hypervisors, making the product attractive to smaller organizations. Scale Computing recently added integrated disaster recovery capabilities into its HC3 platform.
Austin-based NIMBOXX came out of stealth last summer with $12 million in funding, touting its hyperconverged system as the "atomic unit of the software-defined data center." Like Scale Computing, NIMBOXX is KVM-based. The core component of its architecture is its Mesh operating system installed on bare metal. NIMBOXX distinguishes itself from other hyperconverged systems with its home-grown software stack, which includes some open source components such as KVM, John Abbott, founder and distinguished analyst at 451 Research, wrote in a report. The bare metal operation enables "maximum performance without the usual license fees and performance overhead from traditional hypervisors," he said.
VMware jumped into the hyperconverged space with the introduction of its EVO:RAIL platform last summer. EVO:RAIL includes VMware Virtual SAN (VSAN) and serves as a reference architecture for hardware vendors. Companies offering EVO:RAIL systems include HP, Dell, SuperMicro, and Hitachi Data Systems, and EMC recently joined the party with its own EVO:RAIL system, VSPEX Blue. While vendors tout EVO:RAIL and other hyperconverged systems as cheaper than traditional storage systems, the cost of an EVO:RAIL system may be surprising in practice.
Pivot3 is hardly a newbie, founded in 2002 with a focus on converging virtual servers, storage and networks. The company says it launched its first hyperconverged appliance in 2008, when a casino asked for a secure and cost-effective way to store video streams (at the time, Pivot3 called it "serverless computing.") Using Scalar Erasure Coding, Pivot3 developed vSTAC OS, which the company says "allows any program running on one appliance within the cluster to access resources across all the appliances in the cluster." Pivot3 focuses on the video surveillance and virtual desktop markets and counts more than 1,300 customers worldwide. Last month, it closed $45 million in funding.
Maxta takes a software-based approach to hyperconvergence, providing software-defined storage products that can be deployed on x86 servers using any hypervisor and storage device combination. Maxta's MaxDeploy reference architecture offers predefined ways for combining Maxta's software with partner hardware platforms, including Cisco Unified Computing System (UCS) C-Series Rack-Mount systems. Maxta closed $25 million in funding last spring.
Gridstore offers hyperconverged infrastructure products purpose-built for Microsoft Hyper-V. The startup's hyperconverged appliances come in both all-flash and hybrid versions. Unlike other scale-out storage products, which use standard storage protocols such as SMB or iSCSI, Gridstore places much of the work of managing the scale-out cluster into the client as a virtual controller. Gridstore may have an advantage in the market if it can capitalize on its position as the first Hyper-V optimized storage system.
Israeli startup Stratoscale is developing a "hardware-agnostic" software platform that converges compute, storage and networking and supports both virtualized and container workloads. The platform is designed to automatically distribute physical and virtual assets in real time for operational efficiency, and is centrally managed. In November, Stratoscale said it received $32 million in funding, led by Intel Capital.
Bob Friday, Chief AI Officer for Juniper Networks, explains how the advanced technology is transforming operations.
Contact center leaders from 8x8, Awaken Intelligence, and 360insight discuss the importance of agent experience.
AI may force enterprises to rewire parts of their data centers so they are fully optimized to run such workloads. The question is do you use Ethernet or InfiniBand?
