by Jason Bloomberg, President, Intellyx
In spite of their early promise, private clouds have proven to be far more trouble than expected. They are expensive to set up, difficult to configure, and often fail to deliver on the core promises of the cloud, including elasticity and straightforward self-service usage.
Elasticity – the ability to scale up automatically as capacity demands rise, and then scale back down – generally works, that is, until it doesn’t: every private cloud has an all-too-finite physical capacity several orders of magnitude smaller than the theoretical limits of any public cloud.
Furthermore, self-service capabilities are simple to understand but surprisingly difficult to implement and manage. Private cloud infrastructure must be able to support simple provisioning of compute resources (VMs and containers), storage assets, as well as virtual networks seamlessly, across arbitrary numbers of physical services, independent of the underlying hardware and physical network – a challenge that has stumped many vendors.
In fact, with Stratoscale, the cluster assembles itself. If an existing cluster is short on compute or storage, operators simply add additional servers. In fact, it’s possible to add more servers with high storage capacity or large amounts of RAM as needed. Stratoscale automatically adds those resources to the cloud. To Stratoscale, storage, compute, and network are all simply ‘capacity.’
Implementing a Firm User Abstraction
Stratoscale divides the world into two personas: the user, who is the consumer of cloud services, and the operator, who manages the Stratoscale deployment. Unlike many other private cloud solutions, Stratoscale firmly enforces this division, so users never have to worry about the underlying technology nuts and bolts.
The Stratoscale cloud is multi-tenant, providing each user with their own self-service environment separate from other tenants. The platform also includes a ‘single pane of glass’ interface for operators as well as users.
Cloud users can independently provision and configure infrastructure resources, launching VMs, allocating storage, or building their own isolated networks via Stratoscale’s simple user interface.
The UI is not the whole story, however. Stratoscale also provides API-driven control with AWS-compatible APIs, enabling more complex cloud deployment scenarios as well as seamless transitioning to the AWS environment.
Furthermore, Stratoscale handles security, data protection, replication, and high availability automatically behind the scenes. From the user’s perspective, therefore, Stratoscale provides the security and control of a VMware environment, coupled with the agility, time-to-value, and self-service of AWS – the best of both worlds.
For operators, the ‘single pane of glass’ interface simplifies deploying and configuring Stratoscale with existing IT skills. Using this interface, operators can manage multiple customers, projects, or individual instances within the same cluster, leveraging Stratoscale’s strict resource isolation and consumption policies.
Architecture that Gets Hyperconvergence Right
Hyperconvergence is a hot topic in data centers today. It refers to hardware that integrates compute, storage and networking within individual servers, as opposed to dedicating separate hardware for each infrastructure function.
Stratoscale achieves the agility and simplicity of hyperconverged infrastructure with a software-only solution. Every server runs a similar software stack, and together with the other servers forms a private cloud.
Clustering such hyperconverged servers to provide horizontal scalability and elasticity has been a challenge, however – a challenge that Stratoscale addresses for any Intel x86 hardware. Stratoscale’s architecture appears in the diagram below.
Stratoscale Private Cloud Architecture on Hyperconverged Servers (Source: Stratoscale)
Stratoscale provides compute, storage, and network virtualization across hyperconverged clusters. This virtualized infrastructure, in turn, supports a cloud manager and integration layer for centralized control.
Via this straightforward private cloud architecture, Stratoscale simplifies IT processes, thus reducing infrastructure and staffing costs. In fact, Stratoscale’s automatic clustering balances elasticity and cost – allowing the deployment of inexpensive, modestly-sized clouds for proofs of concept and simple dev and test environments that can easily scale up to full enterprise-class production.
The Intellyx Take
Building a private cloud with Stratoscale may sound too easy. After all, many larger vendors have tried and failed to put together a solution as simple and scalable as Stratoscale’s, with no luck.
Look a bit deeper and you’ll soon realize that Stratoscale has put in the time and effort to produce the innovation that this level of technology exhibits. In fact, it has filed more than 20 patents, and have several more in the works.
Many of its patents deal with how clusters of VMs leverage and share memory, including the live migration of VMs that use externalized memory pages, unified caching of storage blocks and memory pages, cluster-wide memory management, and memory resource sharing: enabling a VM on one physical node to use the remote memory pages found on another physical node in the cluster.
Yet, even with its patented innovations, Stratoscale’s accomplishments may strain credulity. There’s one way to convince yourself it actually works: try it for yourself.
Copyright © Intellyx LLC.
Stratoscale is an Intellyx client. At the time of writing, none of the other organizations mentioned in this paper are Intellyx clients. Intellyx retains full editorial control over the content of this paper.