Hortonworks has announced the details for the latest version of its Data Platform product, with a focus on cloud and deep learning workloads, as well as some extended partnerships with two of the big three cloud vendors to help customers running hybrid deployments.
The company made these announcements at the DataWorks Summit in San Jose this week. Chief product officer Arun Murthy stated in a press release: "The pace of innovation coming from the open source community has not slowed and means that customers are getting the latest and best new features in HDP, including containerisation, ability to run deep learning applications and major performance enhancements to analytics."
Hortonworks Data Platform 3.0
The vendor's core Data Platform (HDP) is essentially an enterprise-ready version of the open source Apache Hadoop data platform. In the 3.0 release the vendor is focusing on containerised workloads, support for deep learning use cases and helping customers get closer to real time analytics capabilities.
As well as supporting containers, HDP 3.0 will also allow customers to simplify deployments of complex and compute-intensive machine learning and deep learning workloads on top of their data by allowing for 'pooling' and 'isolation' of GPUs.
In terms of core performance, Hortonworks has also been working on query optimisation to help customers get closer to real time analytics.
By leveraging Apache Hive 3.0, the 3.0 release gives customers "the only unified SQL solution that can perform interactive query at scale – regardless of whether the data lives on-premises or in the cloud," the vendor said.
There is also enhanced security and governance as the vendor helps customers dealing with new responsibilities as a result of GDPR, such as a full chain data custody, data lineage and fine-grained event auditing capabilities.
HDP 3.0 is expected to be generally available in Q3 2018.
As more customers look to run HDP in hybrid environments, Hortonworks is ensuring its platform supports all the major cloud providers, using the San Jose event to announce three new cloud partnerships, all hinging on supporting hybrid deployments.
The expanded Google Cloud partnership hinges on giving customers of HDP and the Hortonworks DataFlow (HDF) platforms more flexibility when running on Google cloud infrastructure, by allowing them to run a hybrid data architecture, where they can move data easily between on-premise and GCP deployments.
The Microsoft partnership is also an extension to an existing relationship which will give customers more choice as to where their analytics and IoT data workloads run.
So customers can deploy any Hortonworks product on Microsoft Azure infrastructure as a service (IaaS), or use Microsoft Azure HDInsight as a fully managed service.
Hortonworks is clearly responding to a strong customer demand with these announcements.
"Our customers are increasingly adopting a hybrid data architecture as cloud deployments offer excellent use cases for ephemeral analytic workloads," as Hortonworks CEO Rob Bearden said.
HDP can be run on Amazon Web Services (AWS) infrastructure but there was clearly nothing to announce regarding this relationship.
Finally, Hortonworks announced an IBM Hosted Analytics solution for customers looking to run a managed service on the IBM Cloud.
As Rob Thomas, GM at IBM Analytics Platform put it in a blog post: "Today, we’re announcing the IBM Hosted Analytics with Hortonworks (IHAH), which presents our integrated solution available as a service on the IBM Cloud. With IHAH, users are given a fully-provisioned data management and analytic environment via the cloud, for very quick setup, provisioning, security and deployment."