Scale-out workloads used to be the domain of high-performance computing (HPC). However, over the past couple of years scale-out workloads have arrived in the enterprise with applications like machine learning or big data/analytics. Enterprise environments, and increasingly HPC users, require 24/7 availability of the storage system.
What is BeeGFS?
BeeGFS is an open-source, parallel file system that was envisioned to be an easier-to-use alternative to common HPC file systems. With a focus on high performance, BeeGFS followed the route of many HPC file systems and relegated redundancy and availability to dated hardware solutions like RAID controllers.
BeeGFS vs. Quobyte: How they compare
Quobyte is a parallel distributed file system that stores data and metadata redundantly across machines to provide scale-out high performance with high availability and automatic failover. Unlike BeeGFS, Quobyte does not require downtime when a machine or disk fails. All tasks like updates or re-configuration are non-disruptive.
In addition, Quobyte is a software-defined storage platform that comes with a range of features not found in BeeGFS. These features enable you to consolidate multiple storage applications on the same cluster (hardware isolation, policy-based placement). Also, features such as multi-tenancy, oversubscription, and tiering, enable you to do storage as a service for multiple groups or departments. Additionally, with Quobyte you can deploy anywhere- any server, public cloud, as well as the cost-effective combination of flash and HDD, even within the same file.
| BeeGFS | Quobyte | |
| Type | Parallel File System | Parallel File System |
| Data protection | hardware RAID within node software RAID-0 across nodes with potential split brain | split-brain safe synchronous replication or erasure coding across machines, racks, or data centers |
| Space efficient erasure coding for files | no | yes |
| Snapshots | no | yes |
| Combination of flash and HDD | only as pools | pools or automatically for small files and within files |
| End-to-end checksums for data integrity | no | yes |
| Clients | Linux | Linux, Windows, macOS Hadoop/HDFS |
| Non-disruptive upgrades | no | yes |
| Non-disruptive data movement (e.g. drain) | no | yes |
| Non-disruptive and automatic rebalancing | no | yes (configurable, not mandatory) |
| Policy-based data management | no | yes |
| Tiering | yes, manual | yes, automatic |
| ACLs | yes | yes |
| Object Storage (S3) | no | yes (same namespace as file system) |
| Multi-tenancy | no | yes |
| End-to-end data encryption | no | yes |
| TLS | no | yes, configurable per network |
| X.509 certificates for authentication with untrusted clients | no | yes |
More Efficient Storage With Quobyte
Quobyte combines high performance with exceptional reliability. You can serve multiple HPC storage types from the same Quobyte cluster, such as home directories and scratch, reducing your cost and storage administration overhead.
With the policy-based data placement, you can easily implement hardware and performance isolation for the different storage types, e.g. to ensure home volumes are not impacted by IO-heavy jobs that access scratch storage.
Unlike BeeGFS, Quobyte provides unified access to files from a range of interfaces including native drivers for Windows, HDFS, and S3. Object/S3 and file system access the same namespace, i.e. a file system can be shared externally via the S3 interface.
Find out how Quobyte scales capacity and performance by downloading and installing the Quobyte Free Edition.
