In the world of Docker and Kubernetes, the notion of plug-and-play infrastructure seems hardly likely. But for organizations trying to reduce time to market, time is of the essence, so packaged appliances may provide a path forward.
In The New Stack’s ebook “The State of the Kubernetes Ecosystem,” research director Lawrence Hecht explored Kubernetes’ reputation for complexity, noting the learning curve gets better over time for the open source container orchestration engine.
Diamanti is focused on the networking and storage aspects of getting up and running fast, especially for large enterprises. So it made sense for the company to release a Kubernetes packaged appliance that, the company boasts, can be up and running in 15 minutes.
“One of the challenges for IT is standing up infrastructure, downloading software and putting it all together,” said Jeffrey Chou, Diamanti CEO and founder. “Our premise is a one-stop shop, one throat to choke. Plug it in and it works. You don’t have to hire a team of engineers to build something.”
Hyperconvergence + Bare Metal
San Jose, Calif-based Diamanti was founded in 2014 by veterans of Cisco, Veritas, and VMware who believed legacy infrastructure such as that from VMware, HP and Cisco didn’t really address the up-and-coming new architectures. Gartner has since named Diamanti among its Cool Vendors for Compute Platforms 2017.
The prepackaged compute, network and storage works well for the legacy data centers many large enterprises still use, he said, though the company might deliver solutions differently down the road.
The appliance, combining hyperconvergence with the performance and efficiency of bare-metal containers, uses standard Linux network interfaces and NVMe block storage volumes that reduce CPU overhead and deliver 100-microsecond read/write latency; 10 Gb Ethernet with configurable, real-time traffic priorities per container extends NVMe across the cluster to offer data mobility.
Setup involves four commands: Create the cluster, set up the network, add persistent storage and deploy. Users can create networks and block storage volumes through the GUI, command line, or flexible REST API.
“We provide a solution that IT can stand up in minutes, can provide developers with self-service containers, and developers can work with their own open source tools without having to learn anything else,” Chou said.
Developers can define infrastructure policies integrated with Kubernetes and Docker to move containers from development to production with predictable application performance.
“A lot of our IP is how we integrate bare-metal containers with network and storage. One of the things on the network side is you have all these overlay networks, with Docker, port maps, VLANS, and so on, and the network administrator needs to create a separate silo for the network. We bring together kind of a Cisco model where it works with any switches you might have. The network administrator doesn’t have to do anything special, we work with all existing network services, Chou said.
It delivers persistent storage with snapshots, mirroring and more.
The company has been involved with the upstream Kubernetes project, contributing Flexvolume, which expands policies and labels you can specify when you create a storage volume for your container.
Diamanti automates control of isolation boundaries to deliver repeatable compute, network, and storage segmentation. Embedded performance queues within the appliance connect every container with their network and storage resources to provide dedicated I/O paths and eliminate noisy neighbors. With role-based access, it can ensure performance unaffected by other developers or other applications running on the same infrastructure.
Its approach to container networking focuses on using existing switches and familiar network services.
One of its customers runs a database-as-a-service offering from its appliance, Chou said. Its customers include NBCUniversal, ShoreTel and MemSQL.
The DIY approach is the company’s biggest competitor, according to Chou. But rolling your own Kubernetes deployment can be expensive and hard to maintain.
“When it comes to open source support, we’re there for you. If you find there’s a problem between the interaction between Docker or Kubernetes and your storage or network; if you cobble it all together, how do you fix this?” he said. “We’re there, a single vendor, and we deliver fixes, patches, upgrades, and we’ll work with the open source upstream … and we’ll support all the way down to the bits and bytes.”
Feature image via Pixabay.
The New Stack is a wholly owned subsidiary of Insight Partners, an investor in the following companies mentioned in this article: Docker.