The cloud may be the new, ultra-convenient data infrastructure, but it must still be provisioned and deployed in order to deliver value to the enterprise.
To do that, IT executives must make a choice not unlike the one they faced in the old bricks-and-mortar days: Either a single-vendor turnkey solution or a piecemeal approach that combines a variety of solutions up and down the stack.
To date, most cloud deployments have been piecemeal, driven either by individual business units creating their own data environments or by centralized IT trying to build private clouds using various vendor solutions. Lately, however, full turnkey platforms have begun to hit the channel, offering ease of deployment but not necessarily the same loss of flexibility that characterized traditional solutions.
IBM recently delivered its first mainframe-based Enterprise Cloud System to UK MSP Vissensa, providing a preconfigured, pretested Linux solution that the company says outperforms x86-based platforms for about half the cost. The package is built on System z hardware, with related storage and cloud management platforms, to provide a complete IaaS solution capable of supporting up to 6,000 virtual machines. Vissensa executives say the mainframe approach allows them to provide a full suite of cloud services from smaller, even mobile, data centers with built-in cost containment capability through programs like IBM’s consumption-based pricing.
HP is joining the turnkey club as well, offering integrated server, storage, networking and software bundles under its CloudSystem program. But rather than simply offering generic solutions for general-purpose workloads, HP also provides a fair amount of customization that allows the enterprise to craft cloud environments to suit unique data and application needs. When Kraft Foods split its North American and overseas operations into separate subsidiaries, for example, HP was able to devise multiple data centers, virtual environments, Microsoft and SAP software stacks and private cloud architectures capable of self-service provisioning and support for more than 2,200 virtual instances.
Enterprises are also turning to multitenant cloud providers to create virtual turnkey environments that cater to specific operational needs. As Karl Strohmeyer, president of the Americas for data services provider Equinix, points out, multitenancy allows enterprise users to draw from a wide range of services and software offerings, often using private network connectivity that bypasses the congestion of the public Internet, which can hamper standard cloud functionality. In this way, the enterprise can build their own “interconnected business ecosystems” within the cloud, while service providers tap into a ready-made pool of prospective clients.
Turnkey solutions aimed at specific workloads are also coming online. For example, Databricks recently launched the Databricks Cloud aimed at providing scalability and advanced cluster management for the open source Apache Spark processing engine for Big Data loads. The platform supports functions like interactive queries, streaming data, machine learning and graph computing using a single API, as well as improved resource provisioning and management. As well, the solution provides built-in applications for data discovery and exploration, dashboard management and job execution.
Turnkey clouds, then, are not like the turnkey infrastructure solutions of old. For one, they are much more dynamic and can be reconfigured to meet changing data needs with relative ease. As well, they often provide more avenues for third-party solutions to enter the fold, particularly if they are built on open source platforms or commodity hardware.
But turnkey solutions still bring with them the single-vendor reliance, which was supposed to be one of the top limitations of traditional infrastructure that the cloud was supposed to remove. This is not necessarily a bad thing, however, given that few organizations have the knowledge or experience to craft world-class cloud environments on their own.
A turnkey system provides a relatively quick introduction to an integrated cloud environment while leaving room for plenty of fine-tuning later.