Startup Verilume wants to put the spare capacity in your data center to work. The company’s software aggregates idle compute and available storage and packages it up for self-service consumption by developers and data scientists within your organization.
The company uses OpenStack, Ceph, and Open vSwitch, along with its own secret sauce, to create multitenant private clouds in enterprise VMware environments.
“We use VMware to create a blob of idle capacity and create an OpenStack cloud on top of that,” said co-founder Mike Feinberg in an interview. “We create a virtual machine with CPU, memory, network, and storage.”
Verilume divides its product into three components: Forecaster, Cloud Builder, and Cloud Service.
Forecaster tracks spare capacity across a near-term timeframe, such as 72 hours. It builds its picture of available resources by importing usage data from VMware and collecting its own metrics. It uses machine learning and statistical modeling to determine the likelihood that assets will be available.
Feinberg cited a simple example: assembling resources to run a Hadoop job from 5:00 p.m. to 5:00 a.m., when typical business systems would otherwise sit idle.
Cloud Builder orchestrates and instantiates the actual cloud environment. As mentioned, it is currently designed for VMware or bare metal, but the company says it plans to integrate AWS by the end of the year.
While the company relies heavily on open source, its intellectual property is built around workload management to track resources and deliver the private cloud. Feinberg claims it can bring up its cloud environment in minutes.
“The whole point of Cloud Builder is to take out the expertise of creating these environments,” he said.
The third component, Cloud Service, is a SaaS portal where developers log in to find available resources and request services. As mentioned, resources can be consumed as a VM, but Verilume can also provision Hadoop clusters.
The notion of working with spare capacity is intriguing, but what happens when an ad hoc project needs to give back resources?
Feinberg said dev and test environments, which are a core focus of the product, tend to be cyclical in nature, so having resources that come and go may not be disruptive. He anticipates customers will use the product for both dynamic and static uses.
Verilume positions its product as easy to use, so I’m curious why the company isn’t just coming to market as a straight-up provider of simple-to-deploy private clouds.
There’s a lot of activity happening around OpenStack to make it more easily consumable in the enterprise, including most recently the launch of a converged system from Mirantis, which bundles the Mirantis OpenStack package with server and storage hardware from Dell and network gear from Juniper.
When I asked Feinberg, he said the company didn’t want to get “pigeon-holed by the industry as a private cloud company.”
Perhaps Verilume has its eye on one day being a hybrid orchestration system that bridges private and public infrastructures, and wants to retain flexibility in its market positioning.
By targeting spare capacity, Verilume differentiates itself from other private cloud vendors. It may be an easier sell if you can tell a customer they can maximize existing infrastructure investments rather than have to buy new gear to embark on a private cloud project.
This creates opportunities for modest wins within the organization, which may be more appealing than the prospect of standing up a dedicated private cloud environment.
The big questions for potential customers are how well Verilume has streamlined the complexity of creating multitenant cloud services within the enterprise, and how good its resource management and scheduling software is.
To help answer those questions, the company will let you try before you buy with free version for 3 nodes per cloud on 2 clouds. An enterprise license requires a 10-node minimum deployment.
The startup was founded in 2013 and is privately backed. Co-founder Mike Feinberg was previously the GM and Senior Vice President of EMC’s Cloud Infrastructure Group. Prior to EMC, he was VP and CTO of Network Storage at HP. He was also a VP at Morgan Stanley.
Co-founder Dan Petrozzo was previously at Goldman Sachs, where he was a partner and Global Head of Technology for Investment Management. Prior to Goldman he was CIO at Fidelity Investments.
Co-founder Rossen Dimitrov was Senior Technologist and Chief Architect for EMC’s Atmos product. He’s also held positions at Calxeda, Verari, and Digital Systems.