The Apache Hadoop distributed computing platform has grown from its initial markets and use cases into a technology on the brink of much wider adoption. Is it time for managed services provider (MSPs) to take notice? A few service providers already have...
Consider the following:
Z Data Inc., a San Diego-based provider of analytics solutions and big data services, offers Hadoop support and managed services. The company provides three tiers of services, delivered remotely or onsite: Level I consists of a Hadoop help desk for handling technical incidents, Level II provides monitoring, tuning, patching, and backups as well as a help desk, and Level III includes the same set of services as Level II and adds 24x7 coverage for critical issues, according to the company.
Global Computer Enterprises Inc., a cloud services provider based in Reston, Va., has harnessed Hadoop to support its offerings, the most recent of which is a litigation support and e-discovery cloud.
Amazon, IBM, and Microsoft offer Hadoop in the cloud. Amazon hosts the framework on EC2 and S3. The company’s service is dubbed Amazon Elastic MapReduce (MapReduce being one of the core components of Hadoop).
In October, IBM announced the arrival of InfoSphere BigInsights, the company’s Hadoop-based analytics software, on its SmartCloud Enterprise infrastructure as a service offering. And Microsoft late last year launched Hadoop-based services for Windows Azure as a community technology preview.
Hadoop: Expanding Opportunity?Hadoop aims squarely at big data problems. The software is designed to process very large sets of data, harnessing clusters of computers to get the job done. Hadoop got its start in search engines and also found use among Web companies in click stream analysis. Hadoop now crops up in other industries that tend to be awash in data: financial services, life sciences and law (e-discovery), for example.
Hadoop distributions are making the technology, if not quite mainstream, at least more readily available. Distribution vendors such as Cloudera, Hortonworks, and MapR Technologies offer tools for managing Hadoop clusters. So for MSPs, creating a Hadoop managed service won’t prove to be an entirely DIY task.
Cloudera Manager, for example, lets administrators deploy and centrally run a Hadoop stack. A free version of the management application supports up to 50 nodes. An enterprise edition of Cloudera Manager, which the company says includes such features as proactive health checks and intelligent log management, is packaged as part of Cloudera’s Cloudera Enterprise subscription.
MapR, meanwhile, offers Control System, which includes graphical provisioning and planning tools, and Heatmap, which provides visibility into the health of a Hadoop cluster. Hortonworks Data Platform includes Apache Ambari, an installation and management system.
Hadoop Channel Partner ProgramsChannel programs could help MSPs move into the Hadoop space. Hortonworks’ Technology Partner Program, which targets service providers, OEMs, and ISVs, offers free or discounted training seats, architecture review sessions, developer support, and joint marketing opportunities. And Cloudera’s Connect Partner Program offers technical and sales training, marketing support, and discounts on Cloudera University courses among other elements.
The big data crunching capabilities of Hadoop will probably far exceed the needs of many MSP customers. But a subset of a provider’s customer base could well benefit. Plusses for getting involved now? For one, few MSPs seem to be involved with Hadoop services at the moment. Second, the Hadoop platform is a blank slate of sorts where a service provider has plenty of room to add value.