A Hadoop cluster is difficult to set up and configure. How do you distribute your services optimally across all the nodes for your workload? How do you account for fast moving and uncertain Big Data requirements which lead to spiky or unknown infrastructure requirements? Hadoop and Cloud vendors realize the growing importance of providing automated provisioning of Hadoop clusters optimized for the customers’ workloads on an elastic infrastructure. On April 13, 2015, Hortonworks announced the acquisition of SequenceIQ, an open source provider of tools for deploying on-demand Hadoop clusters in the cloud. At its Build conference in late April 2015, Microsoft announced its HDFS compatible repository for big data analytic workloads in the cloud.
Why Bare Metal?
Hadoop is ill-suited to traditional virtualization. It is highly demanding on compute, storage and networking. Average server utilization can be very high and there are very large volumes of persistent data. It also requires dedicated networks and fast interconnects. In a nutshell, Hadoop needs a predictable service delivery environment and cannot tolerate “noisy neighbors”. While virtualized Hadoop clusters offered by some of the leading cloud vendors are adequate for development & testing, dedicated, bare-metal infrastructure is best for production environments. A couple of vendors have recently started breaking the “one size fits all” trend for Hadoop on Cloud and have started offering Hadoop clusters on bare metal in the cloud. After a period of offering Hadoop on virtualized clusters in the cloud, Rackspace announced its Hadoop on Bare Metal cloud offering in October 2014. Bigstep, a relative newcomer in this space, also touts the many benefits of Hadoop on Bare Metal in their post titled Feed Your Hadoop with Bare Metal.
SoftLayer, an IBM company is leading the way in offering bare-metal cloud computing via their world-class global infrastructure. BlueMix, IBM’s Platform as a Service (PaaS) offering enables organizations and developers to quickly and easily create, deploy, and manage applications on the cloud. IBM BigInsights enhances open source Hadoop with a complete set of capabilities to query, visualize, explore data and conduct distributed machine learning at scale resulting in deeper insight and better actions. Put SoftLayer, BlueMix & BigInsights together and you get IBM’s Hadoop as a Service. IBM’s BigInsights on the Cloud offers the rich features of BigInsights via BlueMix on SoftLayer’s bare-metal infrastructure without the cost, complexity and risk of managing your own infrastructure. The dedicated, bare-metal infrastructure provides enhanced performance and security. Each customer’s cluster is isolated on a private VLAN and only select ports on management nodes are accessible to the public network and protected via authentication.
Choose the Big Data category in the BlueMix Catalog and order your Hadoop cluster today.