Technology FAQs

Three technology components make up the Kodiak infrastructure:

A.   An user-space linux ‘fabric node’ (Fabric-node)

It runs on any physical or cloud server that wants to provide local assets (cpu, ram, flash, disk, file-system, mount points etc) for use by the Kodiak cloud fabric.

B.   A client-side linux loadable kernel module (‘mod-kodiak’)

Provides vDisks (kernel virtual disks) to local linux apps. These vDisks are used to access data and storage assets on a Kodiak Cloud. We support various linux kernels starting with REL/Centos 6. This code is open-sourced and Kodiak technology partners are working on porting this to other OS.

C.   A client-side node.js app (Kodiak ‘dockport’)

This provides a management and api hook to run provisioning and authorization on the client side.

These module descriptions are for architectural reference only. They are not visible for Kodiak Data cloud services like MemCloud or StorageCloud – and typically user code does not interact with them.

The Kodiak Fabric nodes runs on top of linux. The northbound API is based of vDisk and virtual network interfaces (optional).

We can run in:

• Hyperconverged – Each server runs donor and client.  Almost 90% of our cloud servers run like this.
• Converged – Clients can run on any node, including diskless/flashless nodes.
• Standalone – Stand-alone data services cloud where user code runs separate and access data services via storage, database, SQL and or NoSQL API.

StorageCloud runs as a standalone or converged cloud. MemCloud always runs converged or hyperconverged.

Two ‘magic sauces’ actually.

• The first one is the Data Fabric

It’s a pioneering approach that allows single and multi-tenant cloud users to scale and harvest the networking and some of the new advances in data infrastructure.

Developed and perfected over a period of two years – and backed by two patents and over a dozen more in the docket.

Contact us and we will be happy to tell you all about it.

• The second part is our Datacontainer Compiler Technology

It allows us to define a clustered, multi-tenant, multi-stack complex infrastructure in a very simple manifest form called Datacontainer™.

The datacontainer compiler then composes a logical infrastructure (VM, docker containers, disks, virtual racks, networking, namespace, policies, lambda components) on our fabric.

In under five minutes.  Magical.  Just ask us for a demo.

We’re API first. All REST API.

We have cloud-based portals and dashboard services that ALL use the API and Oauth in the back-side.

The central cloud-based dashboard is CloudWorks.

CLI? Seriously?

MemCloud is hosted in our own data-centers and in-building data closets. We have an expanding set of locations in SF bay area (primarily for diversity). Contact us for current locations.

We offer our MemCloud in AWS. You won’t get the same performance or savings of MemCloud running on Kodiak Data Centers. But you will get the features, functionality, convenience, ease of deployment of Kodiak Cloud. Currently it’s invite-only and for medium to large enterprises.

Contact us to learn more.

All our dashboards, portals and other services like Cloudworks run in Google Cloud.

StorageCloud:  Yes. StorageCloud is available under a SW-only license for standard servers.

MemCloud:  Yes. This is a custom engagement – contact us if interested.

Note that we do have recommended configuration specs for Dell/HPE/Cisco/IBM and likes. We do not support configurations with only hard disks or with networking speeds of less than 10GE per server.