Oracle Cloud Infrastructure HPC File Systems (HFS)

High performance file systems support workloads that require the ability to read and write data at extremely high throughput rates. OCI HPC File Systems (HFS) are available on Oracle Cloud Marketplace and make it easy to deploy a variety of industry leading high-performance file servers. In as little as three clicks, customers have a file server up and running at petabyte scale with double digit gigabyte throughput.

When you deploy these HFS solutions

When you deploy these HFS solutions, you can specify the requirements for your file system, including:

  • Type (scratch or persistent)
  • Workload Type (small, mixed or large files)
  • Filesystem Sizing (number of file servers and storage capacity)

As you specify these parameters, the HFS Marketplace solution then applies default parameters to reduce customer complexity and shorten time to deployment. For example, when building a scratch filesystem, it will select local NVMe storage. Similarly, for mixed or small file workloads it will customize metadata. File systems can be up and running in under 15 minutes.

Using HFS Marketplace solution, in just a few clicks you can deploy BeeGFS, Lustre, and GlusterFS.

HFS Customizable Options

For users who require maximum configurability for complex deployment, we also offer Oracle Cloud Marketplace Stacks (Web based GUI) and Terraform-based Oracle quickstart deployment templates on Github. Simply customize the terraform templates or Marketplace stacks to meet your requirements and launch through Resource Manager.

Benefits of HFS on OCI HPC

Built on Oracle Cloud Infrastructure, these shared file storage solutions provide high throughput and consistent, low latency, supporting high performance workloads.

Fast

Achieve file throughput of up to 500 GB/s aggregate throughput with attached HPC bare metal compute instances with local NVMe and 100 Gbps cluster networking for low latency.

Easy to Use

Get up and running quickly and easily, with just a few clicks to deploy standard configurations.

Everything You Need

The all-in-one solutions based on Terraform scripts include everything you need to get POSIX -compliant Lustre, GlusterFS, or BeeGFS up and running quickly and easily.

Low Cost

On-premises file server clusters require large capital investments, while Oracle Cloud can scale elastically to meet storage and performance needs up to 500 GB per second.

Use cases

Oracle’s preconfigured environments are useful in many industries across a wide range of applications, including:

  • Data migration to cloud

    HFS can be used to build a POSIX file system that’s accessible from on-premises compute nodes using file system native clients, NFSv3/NFSv4, or SMB to migrate data to Oracle Cloud.

  • Engineering and scientific computing

    HPC workloads for crash simulations, computational fluid dynamics (CFD), and computational chemistry require massive amounts of training data and often use shared file storage to support processing data sets concurrently.

  • Artificial intelligence and machine learning (ML)

    Data sets for AI and ML workloads are large, and file storage must supply massive amounts of data very quickly for training and validation across multiple models simultaneously. OCI supports WEKA, Lustre, GlusterFS, and more.

  • Media processing and transcoding

    Media data processing workflows, like video rendering, visual effects, and media production, need the compute and storage resources to handle the creation of massive amounts of video files from large resource data sets.

  • Big data and financial analytics

    Big data analytics use cases, including fraud detection and financial analysis, produce massive volumes of data that need high performance storage to power data-intensive applications.

"Oracle Cloud is the only major cloud provider delivering Spectrum Scale in a shared-storage architecture on bare-metal hardware. The Oracle offering has been tuned for the demands of HPC and big-data analytics and offers performance advantage over virtualized cloud offerings, improved flexibility, scalability, and manageability."

Michael Sedlmayer President, Re-Store

“My team has tested SAS Grid on many public clouds. We are happy to say that Oracle Cloud’s infrastructure was able to provide the I/O throughput to the IBM Spectrum Scale shared file system that is needed for SAS Grid.”

Margaret Crevar Sr Manager, SAS Performance Lab, SAS

“The OCI HPC File system (HFS) graphical interface and its file system provisioning workflow, designed to be customer use case centric (scratch or persistent space, large, mixed or small file workloads, etc.) versus technology centric, reduces barriers to quickly deploy a parallel filesystem like BeeGFS on Oracle Cloud. Oracle Cloud offers infrastructure resources like 100Gbps RDMA RoCEv2 Clustered Network, Multi-attach Shared Block Volume service, local NVMe SSDs attached to Compute & Bare metal Compute shapes to deploy BeeGFS in many different configurations and deliver triple digit Giga-Byte throughput (500 GB/s) at lower price point.”

Howard Weiss Founder and Managing Director, Pacific Teck

“Oracle's bare metal compute and clustered networking technology allowed BeeGFS on Oracle Cloud to outperform on-premise HPC file system performance (latency & IO throughput) for MPAS workload at a very low price point. Using Oracle Cloud's Cluster Network RDMA network, BeeGFS can see performance of up to 140 GB/s with as little as 14 servers.”

Simon Ponsford CTO, YellowDog

“The OCI parallel file system solution of BeeGFS using 100 Gbps RoCEv2 and local NVMe storage achieved similar or better IO write throughput performance for the SWIFT cosmological application benchmark depending on the file type written. For 370 GB snapshot files, OCI achieved 6.2 GB/s and Lustre 7.0 GB/s and for 994 GB restart files OCI achieved 70 GB/s and Lustre 29.2 GB/s.”

Andrew Turner DiRAC Technical Manager, EPCC, The University of Edinburgh

HPC Filesystem on OCI – Feature Comparison

Features BeeGFS GlusterFS IBM Spectrum Scale Lustre
License Type Open Source & Commercially licensed for Enterprise features Open Source Commercially licensed Open Source
Scalable - Storage & Performance Yes Yes Yes Yes
Replication Yes* Yes Yes Yes
High Availability Yes Yes Yes Yes
Storage Tiering Yes* No, experimental Yes Yes
POSIX Yes Yes Yes Yes
NFS Version v4 v3, v4 v3, v4 v3, v4
Server Message Block (SMB) Yes Yes Yes No
RDMA Support Yes No Yes, Erasure Edition No
Filesystem Support SI & Community Community SI SI & Community
Deployment HFS, Oracle Quickstart HFS, Oracle Quickstart Oracle Quickstart HFS, Oracle Quickstart

* - Enterprise feature, available in commercially licensed software. Open Source edition include all enterprise feature to evaluate them for free for non-production use.
SI – stands for System Integrator

April 15, 2022

WEKA on OCI delivers 2 terabytes per second performance

Pinkesh Valdria, Principal Solutions Architect

Oracle and WEKA are proud to announce the availability of WEKA on Oracle Cloud Infrastructure (OCI). Both companies have been working together to validate that WEKA can run on OCI bare metal instances to deliver the highest level of performance that customers demand for their critical workloads. Along with performance, we’ve also validated transparent storage tiering for WEKA by using S3-compatible OCI Object Storage as a low-cost tier. This option enables customers to run at petabyte scale in a high performance file system by using local NVMe SSDs for hot data and object storage for warm or cold data.

Read the complete post

Resources

Get started with Oracle Cloud


Oracle Cloud Free Tier

Build, test, and deploy applications on Oracle Cloud—for free. Sign up once, get access to two free offers.


Cloud Training—Oracle Cloud Infrastructure

Explore cloud training resources with Oracle Cloud Infrastructure training videos, self-paced learning labs, and certifications.


Explore Oracle Cloud Infrastructure

Oracle Cloud Infrastructure combines the elasticity and utility of the public cloud with the control, security, performance, and predictability of on-premises computing environments.