HPE Ezmeral: Uncut

HPE Ezmeral simplifies multi-cloud management and protection

HPE_Ezmeral_data-protection-movement.pngEver since disk drives replaced tapes for storing business critical data, business continuity and disaster recovery have been key elements for any size enterprise. Business continuity ensures operations continue even during unplanned events, such as a fire in a branch office. Disaster recovery by its name presumes that business ops have been impacted and IT infrastructure and operations need to be restored as quickly as possible. In either use case, I have no doubt appliance or cloud-based snapshots and mirrors are the primary mechanisms used to restore business operations. Depending on the business vertical, business continuity plans and testing are required to protect CRM, order systems, and customer data critical to everyday business. 

But if you are like most companies, hundreds of petabytes of analytic and AI/ML data could be exposed to natural or man-made events that restrict data access, slow insights and innovation, or even impact the competitiveness of your business. If you are using traditional solutions for analytic and AI/ML use cases, are you getting the cost-efficiency and performance you need? How long does it take to backup and recover 5-, 10-, or 40-petabytes of data with appliance or cloud-based solutions?

HPE Ezmeral Data Fabric

HPE Ezmeral Data Fabric is unique in the industry with a software-defined data store that accelerates the ingestion, processing, analyzing, and storage of all data types from any source, infrastructure, or multiple ingest mechanisms. It integrates complex data structures and sets across core, cloud, and edge into a single source of truth that supports smarter, more certain business decisions to speed business growth. This proven solution comes with built-in snapshots and mirroring to not only ensure data and applications are protected, but can also move data from on-premise to cloud, cloud to cloud, or transparently tier data into object stores without losing metadata.   

Data Fabric Volumes

To understand how HPE Ezmeral Data Fabric works, it is important to understand a core concept to the solution:  the data fabric volume.  

Data fabric volumes are logical file system structures that hold directories, files, tables, and streams. They ensure platform-level orchestration through a single policy that spans multiple items and are responsible for defining properties, replication, erasure coding, security, data access and root/administrative permissions (Figure 1).

Figure 1 Simple point and click volume creationFigure 1 Simple point and click volume creation

Data Fabric Snapshots

Figure 2  Protections when restoring  snapshotsFigure 2 Protections when restoring snapshotsSnapshots are point-in-time, read-only copies that follow a redirect-on-write format to deliver storage efficiency. This means when changes occur, the HPE Ezmeral Data Fabric file system writes only the change data and does so without impacting performance.

Unlike standard HDFS, HPE Ezmeral Data Fabric snapshots can be created on demand or automated with a schedule to deliver atomic, point-in-time volume images. Users can restore individual files or roll back an entire snapshot to a specific point in time. When restoring the full snapshot, the data fabric volume must be unmounted before the snapshot is recovered and then remounted as protection against restoration being a random event (Figure 2). 






Beyond data protection, HPE Ezmeral Data Fabric snapshots can be used by engineering and data science professionals for the following use cases. Figure 3 Engineering and data science use casesFigure 3 Engineering and data science use casesData Fabric Mirrors

HPE Ezmeral Data Fabric mirrors are read-only copies of any volume located on the same or different cluster.  Because the platform is software defined, server sizes do not need to be the same; however, if you a replicating 5 Tbs of data, the restore site needs to have 5Tb available capacity. Mirrors update incrementally in 8K blocks and compress to protect network performance. If the network is slow for any reason, replication will slow down or back off until network performance improves. Mirror Fast Safe Flexible.png

Here are some sample use cases for HPE Ezmeral Data Fabric mirroring:

- Researchers, developers, and analysts experiment without impacting the production cluster by mirroring from production to research clusters over the LAN

- Mirroring between data centers over a Wide Area Network (WAN)

- Mirroring from a production on-premises cluster to a HPE Ezmeral Data Fabric cluster running in the cloud


HPE Ezmeral Data Fabric removes the need to mirror the entire cluster allowing users to be highly selective and mirror only the volumes or directories they need. Incremental mirror refreshes can be scheduled with a low recovery point objective (RPO) to minimize data loss or a low recovery time objective (RTO) to minimize time until operations resume.

Network Interruptions  
Any network interruption is handled gracefully. For example, in the event of a network partition, the mirroring operation will periodically retry the connection. Once the network is restored, the mirroring operation automatically resumes without third-party software.

Data Movement & Application Portability
A key issue for organizations today is the ability to easily move data or port applications from on premises to multiple cloud environments. HPE Ezmeral Data Fabric’s support of standard APIs and single data infrastructure allows applications to move across core, cloud, and edge environments and natively synchronize multiple clusters regardless of their location. The solution can also move applications between locations to best match the application to the required environment.  

DF Application Portability.png

With HPE Ezmeral Data Fabric, customers can use mirrors to:

  • Integrate with cloud-based container services on AWS, Microsoft Azure,
    and Google Cloud
  • Meet high availability requirements with data synchronization across cloud availability zones
  • Cross cloud data bursting to support cloud neutral deployments  
  • Easily go from on-premises and private cloud deployments to public cloud

Read the eBook, Simplify Real-Time Data Access, to see how two real-world customers are using HPE Ezmeral Data Fabric to unify and simplify data management.

When it comes to protecting your analytic and AI/ML data, you need to modernize your approach -- you need HPE Ezmeral Data Fabric. This solution delivers a single platform that ingests, processes, and analyzes data as well as ensuring its security and resilience. Your company benefits from a single solution that consolidates management panes of glass and hardware while delivering the multiple cloud data movement your developers and data science teams need. To learn more about how HPE Ezmeral Data Fabric works, read the technical paper, HPE Ezmeral Data Fabric: Modern infrastructure for data storage and management.

Joann Starke

Hewlett Packard Enterprise


0 Kudos
About the Author


Joann’s domain knowledge and technical expertise have contributed to the development and marketing of cloud, analytics, and automation solutions. She holds a B.S. in marketing and computer science. Currently she is the subject matter expert for HPE Ezmeral Data Fabric and HPE Ezmeral Unified Analytics.