Cool Vendors for Compute Platforms, 2016

The rapid adoption of containers and "software defined" abstraction from hardware have created renewed interest in server and virtualization technology. The following Cool Vendors seek to guide I&O leaders how to assess and expand server functionality beyond traditional OSs and hypervisors.

Key Findings

• All five vendors profiled in this research are selling solutions that orchestrate resources and deliver an abstracted service.
• Massive market hype and growing enterprise adoption of hyperconvergence (hyperconverged integrated system [HCIS]) and software-defined storage (SDS) have been a powerful magnet for entrepreneurs and investors who continue to target the market.
• The increasingly vibrant and congested HCIS/SDS market creates challenges for startups to truly differentiate themselves versus rivals, which is resulting in greater scrutiny and declining enthusiasm among the investment community.
• The focus for most new HCIS/SDS startups is shifting from a complete appliance-type hardware/software solution toward a software-only focus, where most seek close collaboration with one or more system vendors with established channels and end-user influence.

Recommendations

• Look to vendors, such as those included in this research, whose products aim to offer simplified and service-focused solutions when staff time and expertise are limited.
• Pressure all HCIS/SDS vendors (startups and more established players) to prove that claimed cost of ownership and functional benefits are viable and compelling, as proof is harder to demonstrate in a congested market.

Analysis

This research does not constitute an exhaustive list of vendors in any given technology area, but rather is designed to highlight interesting, new and innovative vendors, products and services. Gartner disclaims all warranties, express or implied, with respect to this research, including any warranties of merchantability or fitness for a particular purpose.

This document was revised on 5 May 2016. The document you are viewing is the corrected version. For more information, see the Corrections page on gartner.com.

What You Need to Know

In this research, we look at five vendors that are challenging the hegemony of traditional compute platforms:

• HTBase: HTBase sells a range of hyperconverged solutions that are able to support multiple hypervisor layers, plus containers and access to services like Microsoft Azure and Amazon Web Services.
• Nimbix: This company offers high-performance computing (HPC) as-a-service capabilities to enterprises that desire the ability to run HPC workloads without the need for capital or organizational investment.
• Springpath: Springpath works with multiple hardware vendors to create a hyperconverged infrastructure that can support multiple hypervisors and OpenStack.
• StarWind Software: The StarWind software-defined storage solution aims to turn any Windows-based server into a resilient iSCSI-based SAN that can support multiple hypervisors.
• Yottabyte: Yottabyte is a hyperconverged infrastructure vendor that enables enterprises to choose a reference architecture approach or deploy their own hardware.

These technology leaders are developing innovative compute solutions that dovetail well with parallel technology investment areas such as cloud infrastructure, cloud management platforms and storage. They augment and enrich trends like hyperconvergence and software-defined anything (SDx). In doing so, these vendors further blur the lines between traditional server hardware or software, and technology developments that create bridges between servers and other technology classes. I&O leaders should examine these Cool Vendors closely and leverage the opportunities that they provide.

HTBase

Toronto (www.new.htbase.com)

Analysis by George Weiss

Why Cool: HTBase differentiates itself in the hyperconvergence space by delivering on the concept of a software-defined data center (SDDC) infrastructure, while enabling IT to retain existing infrastructure or develop new hardware platforms in greenfields. I&O leaders and decision makers can, therefore, retain and adapt their existing x86 hardware and storage for hybrid cloud solutions, while delivering IT as a service (ITaaS) to their business units. To enable such transformation, HTBase comprises three software pillars: HTVCenter, Fortis and Nephos. HTVCenter centrally manages multiple hypervisors, including open-source kernel-based virtual machine (KVM) and Xen; enables existing VMs to migrate to KVMs; and supports Docker, bare metal, and integration to AWS and Azure. It also manages storage, network and compute resources, with capacity management, storage resilience, high availability, striping, deduplication, snapshot/replication (with resulting disaster recovery) and automation tools through REST APIs.

Challenges: HTBase's marketing has thus far been relatively low-profile as it builds credibility in client accounts with major testimonials. To gain strong customer acceptance and wins, it has been performing free studies and proofs of concept (POCs), and has consulted on SDDC implementations, which require time and investment. Thus, market recognition – in the sense of a volume-based business like Nutanix and SimpliVity – is still elusive on its way toward more ambitious goals of ITaaS, rather than merely specific use cases like virtual desktop infrastructure (VDI) and appliance wins among smaller users. To be a recognized industry success, HTBase must show that its ability to enable implementations of an SDDC is at the cusp of leadership and innovation, with accelerating adoption rates and market interest. HTBase must also prove – together with like-minded vendors – that a "bring your own hardware" policy can still result in predictable performance and viable support for integrated hardware and applications.

Who Should Care: CIOs and I&O leaders designing infrastructure as a service with a strong cloud strategy will want to engage in a dialogue with HTBase, and see demonstrations of its three main pillars: HTVCenter, Fortis and Nephos. Moreover, I&O leaders should investigate migration options from legacy environments (regardless of hypervisor or bare metal) to centrally managed – but independently run – business units, controlling their own resource allocation by either conventional VMs or by containers.

Nimbix

Dallas, Texas (www.nimbix.net)

Analysis by Andrew Butler

Why Cool: Nimbix has identified a largely untapped market for cloud-based high-performance computing infrastructure. Nimbix cloud technology enables companies to leverage a range of HPC workloads, with time-based access payable as a service. Application categories include computer-aided design (CAD), fluid dynamics, structural analysis, finite element analysis and simulations. With offices in Dallas and Houston, Nimbix was formed in 2010 and has a natural addressable market in the local petrochemical industry. Recent Series B funding has allowed the company to expand its data center facilities to a second Dallas location, and it is investing in the necessary infrastructure to target multiple other industries including life sciences, 3D media, manufacturing and financial services. The Nimbix infrastructure supports both CentOS and Ubuntu Linux workloads, with full access to the underlying compute and storage hardware. A large range of interfaces and developer tools are intended to save companies the need to invest in their own license management. By offering HPC "as a service," Nimbix enables companies to execute HPC workloads without the need to invest in expensive and specialized hardware, and the necessary facilities to house it.

Challenges: With only two data center locations – both in Texas – access to the Nimbix infrastructure is gated by geographic factors like cost-efficient bandwidth, plus the common geopolitical considerations regarding data location and access. The company will be challenged to build long-term loyalties, as users will access the Nimbix solution on a pay-as-you-go basis, typically project by project and with much access driven by the line-of-business budget holders rather than by the corporate IT organization. The portal-based access provides clear appeal for pilots and small production workloads. But Nimbix maintains that it has signed up a number of enterprises that run large, complex production workloads that take advantage of the comprehensive API that is built into Jarvice. The portal is intended to make it very easy for enterprise end users to consume their application of choice as SaaS. Users should nevertheless validate Nimbix's ability to support sufficiently large cluster sizes and more complex workloads, to avoid users moving these larger workloads in-house or favoring other cloud solution providers.

Who Should Care: Nimbix's Jarvice HPC-as-a-service concept allows companies to trial HPC workloads very easily, and at modest cost with no capital outlay. The service will appeal to smaller enterprises that cannot afford or justify the investment in HPC hardware, plus larger enterprises that may only require occasional access, or that want to trial an HPC solution before investing in their own infrastructure. The Nimbix solution is accessible to any type of company or geographic market, and Nimbix estimates that international clients make up about around 30% of its portfolio. But the "sweet spot" for most business will be U.S.-based companies that are able to access the Nimbix infrastructure at realistic cost, bandwidth, latency and regulatory validity.

Springpath

Sunnyvale, California (www.springpathinc.com)

Analysis by George Weiss and Andrew Butler

Why Cool: Springpath has created a homegrown distributed file system that unites and optimizes data allocation and performance with a controller running on each node, with the aim of integrating and optimizing data center silos at the data layer. The software stripes data across all nodes of distributed hardware with no requirements for data locality across multiple tiers of memory, solid-state drive (SSD) cache and spinning drives, to maximize performance. Metrics collected and analyzed continually inform administrators how to optimize system performance. Springpath makes bold claims for its analytics and data management – promising "no compromise" performance with snapshots, clones, deduplication and compression at superior performance to other HCIS and software-defined vendors. Another layer of software integrates with hypervisor management (multiple choices), and lets administrators use existing VM management consoles, as well as containers or bare metal. Its APIs enable plug-in capability to OpenStack, containers and third-party management tools.

The portability of Springpath's solution, which enables support for existing legacy x86 infrastructure, is a compelling economic proposition. It enables users with hardware OEM relationships in place (such as with Cisco, Dell, HP, Lenovo and others) to deploy the solution with no additional hardware purchase. With scalability projected linearly to hundreds of nodes and the software offered by a subscription (as opposed to a perpetual license), I&O leaders can license by node counts and independently scale compute and storage nodes according to workload demands. Springpath was founded in 2012 by former VMware lead designers. In 1Q16, Springpath closed a partnership with Cisco that enables Cisco to offer the Springpath hyperconvergence software integrated with its Unified Computing System (UCS) on selected configurations.

Challenges: Increasingly, other vendors are using a software-defined hyperconvergence approach that enables portability among OEM hardware. Competitive gains may compel Springpath into detailed performance and functional comparisons, as well as POCs against other software-based HCIS solutions, demanding time and effort to gain market share. The Springpath solution has gained significant market credibility and awareness through an OEM partnership with Cisco. Cisco's new HyperFlex strategy is built on Springpath software coupled with Cisco-based compute, solid-state storage and switches. But this could prove to be a double-edged sword. Industry rumors have raised the prospect of a Cisco acquisition of Springpath. This is denied by Cisco, although Cisco is known to hold an equity stake in Springpath. An acquisition would change the complexion of Springpath's roadmap and market focus away from independence to dedicated hardware appliance solutions. A Cisco hardware solution, as a new project, would also place additional demands to optimize Springpath to Cisco's Application Centric Infrastructure (ACI). To maintain market objectivity, Springpath must demonstrate an equal level of independence through ongoing third-party hardware support (or alternative OEM relationships), continuing to optimize its solution to varying hardware configurations and creating an efficient reseller channel in both software-driven as well as appliance-based sales. Without this, Springpath will become an effective satellite of Cisco, regardless of its nominal independence or neutrality.

Who Should Care: IT directors and application developers who have existing tuned and optimized hardware who want to continue existing OEM contracts and relationships will be attracted to the flexibility of a portable underlying hardware infrastructure. Users and developers may be interested in HCIS applied to Mode 2-type applications at minimal cost and risk through a Springpath subscription program and through its integration with OpenStack, containers and public cloud.

StarWind

Middleton, Massachusetts (www.starwindsoftware.com)

Analysis by Philip Dawson

Why Cool: StarWind promotes hyperconverged and storage appliances targeting small or midsize businesses (SMBs) and remote offices/branch offices (ROBOs) for high-availability deployments. This element of software abstraction of storage functions and hardware appliance is a form factor increasingly attractive to non-data-center locations, and for extending VM clusters outside of the data center. StarWind originated in a corporation founded in 2003 under the "Rocket Division Software" name. StarWind as we know it was formed as a spin-off in 2008, and adopted the strategy the company promotes today. StarWind was an early pioneer in the market that has emerged for software-defined storage and hyperconverged appliances – with some very early HCIS deployments dating back to 2008. StarWind's main three products include a software-based virtual SAN (similar to VMware's VSAN and the upcoming Microsoft Storage Spaces Direct offering from Windows Server 2016), hyperconverged and storage appliances, and a virtual tape library, all built with StarWind Virtual SAN at the core. Services include an interesting turnkey leasing capability and a platform as a service (PaaS) deployment option helping bridge the compute platform with the cloud infrastructure and packaging in public cloud (for example, Microsoft Azure, where virtual StarWind instances can be deployed). To stimulate interest, StarWind offers online demos, free products and downloads. This ability to "try before you buy" helps to build a strong community around the company and its products, and to provide feedback that helps position StarWind against open-source software (OSS) alternatives. This will appeal to those creating test/development environments and preproduction rollouts, and can help balance the cost considerations of OSS alternatives, especially for remotely managed infrastructure appliances. StarWind has technical partnerships in place, including with Microsoft, VMware, Veeam, Mellanox Technologies, Intel and Dell. This firmly places it as a Mode 1 player, virtualizing storage differently from traditional players, but less so a Mode 2 vendor as there are many OSS alternatives (see Note 1 for an explanation of bimodal IT). With active investments in DRAM caching, in-memory computing and a proprietary log-structured file system with an in-line deduplication, StarWind delivers solutions that are analogous to vendors that are bridging the gap between SDS and the broader hyperconverged appliance space.

Challenges: Focusing on the storage revirtualization market gives StarWind an interesting Mode 1 opportunity. The momentum of Mode 2 containers (rather than hypervisors) provides StarWind with extra potential to package its solutions against existing IT investments – like traditional storage and server vendors – although this is a nascent market opportunity today. It is also trying to carve out a niche for a storage SDS/HCIS hybrid positioned against integrated systems hyperconverged vendors like Nutanix and SimpliVity whose VDI use and ROBO base also have SMB appeal. Finally, the operating environments of OS and virtualization vendors are increasingly creating layered storage functions from the SAN into the OS or VM. This will continue further for the container hosts for Mode 2 deployments.

Who Should Care: Midmarket I&O leaders who desire a move away from SAN, or users deploying ROBO workloads that are not connected at the storage tier, can consider StarWind for remote deployments with commodity hardware and value-driven software. But be cautious on quick payback schedules and look for competitive take outs and acquisitions.

Yottabyte

Bloomfield Township, Michigan (www.yottabyte.com)

Analysis by Andrew Butler

Why Cool: Yottabyte is a software vendor addressing the emerging hyperconvergence (HCIS) and software-defined infrastructure (SDI) markets, whose yCenter and YottaBlox products are marketed as a scale-out, rack-scale solution capable of growing to thousands of nodes – considerably more than the scaling claims of all other HCIS vendors. yCenter is marketed as a fully integrated SDI platform, with a KVM hypervisor built in to avoid hypervisor licensing fees. yCenter is based around Yottabyte's own distributed file system, and can be deployed by service providers on suitable x86 hardware. Yottabyte has also developed an Intel-based reference architecture, with the ability to sell a complete hardware/software solution named YottaBlox. Hybrid cloud capability is built-in, and any YottaBlox running virtual data centers may instantly connect or replicate to any Yottabyte-powered public cloud. Yottabyte offers its own public cloud computing and storage services from Michigan-based data centers today. yCenter introduces a multivirtual data center model on shared infrastructure that completely abstracts all elements of the virtual data center (storage, compute and network) from the underlying hardware. The YottaBlox enterprise start kit is a four-node, all-flash storage-based cluster that supports erasure coding to sustain the loss of up to any four drives across the cluster (or one complete node). Unusual for an HCIS vendor, Yottabyte has also developed its own top-of-rack SDN fabric switches, which integrate fully with Yottabyte's Linux-based yCenter management framework. The company offerings include a scale-out SDS solution. Yottabyte's VSAN supports a broad range of data management and resilience features, including synchronous and asynchronous replication, deduplication, silent corruption detection and repair, data tiering, and snapshots.

Challenges: Yottabyte's business model is dependent on the use of the KVM hypervisor to gain operational cost benefits through the ability to save licensing fees with vendors like VMware. This is a viable model for users who have no VMware strategy today. By embedding the KVM hypervisor within the yCenter management console, and automating most day-to-day functions, Yottabyte is able to shield nonexpert users from most of the complexity that a community hypervisor would usually impose. Yottabyte is not the only vendor making bold scaling claims, and users must verify that failure domain isolation and internode traffic will not make such scaling impossible to achieve. Users must also ensure that existing hardware vendors are willing to support their hardware running in such a configuration.

Who Should Care: Yottabyte's products and services are appealing to users who want a highly cost-efficient HCIS and/or SDI solution; who are not dependent on VMware support; and who are looking for an alternative to building an OpenStack framework. The YottaBlox fully configured SKU is very new, but will appeal to service providers and enterprises who want to deploy an HCIS architecture with very broad theoretical scaling.

Where Are They Now?

Nimboxx

Austin, Texas (www.nimboxx.com)

Analysis by Andrew Butler

Profiled in "Cool Vendors in Servers and Virtualization, 2015"

Why Cool Then: Two vendors – Nutanix (depicted in a 2013 Cool Vendor report) and SimpliVity – have dominated most enterprise interest in hyperconvergence during the past two years. This forces other HCIS vendors to seek new selling propositions in order to get noticed. All HCIS vendors typically address VDI opportunities, and Nimboxx acquired a VDI solution named Verde that was integrated with its technology to create a complete VDI appliance. Rather than pursue a VMware support strategy, Nimboxx also targeted emerging Mode 2 next-generation workloads, such as hybrid and private cloud, and big data analytics, where dependency on legacy software support is much lower.

Where They Are Now: Unfortunately, not all Cool Vendors succeed, and Nimboxx ceased operations in late 2015. At the time of writing, we do not know what will happen to the assets of the company, or how the ongoing support needs of the customer base will be maintained.

Who Should Care: The failure of Nimboxx was predictable to a degree; we believe that the great majority of HCIS vendors who are addressing a market that is close to overheating will ultimately become acquired, or run out of ideas or funding. But many of those vendors will achieve enough market impact to move the market forward in terms of innovation and customer satisfaction. The speed of Nimboxx's failure was a shock to many, especially as the acquisition of the Verde software stack had happened only months before. There is a lesson for HCIS buyers and fellow vendors that a privately owned company can shift from a seemingly viable position to complete closure in a very short time.