Products · Storage
BeeGFS parallel file system.
As a Platinum BeeGFS partner, we design, integrate and support BeeGFS parallel storage on the clusters we build and on yours.
Platinum BeeGFS partner
BeeGFS expertise, on any cluster.
BeeGFS is developed by ThinkParQ. We’re one of their top-level partners in the United States, and we put it to work: designing, integrating and supporting BeeGFS storage on clusters we build, and on clusters built by others.
Design and sizing
Storage servers, metadata, networking and storage pools sized around your workloads, your data and your budget.
Integration
BeeGFS built into a new cluster from day one, or added to the cluster you already have, whoever built it.
Support
Help from engineers who deploy and run BeeGFS every day, for the systems we install and the ones we inherit.
Expansion
Add capacity or performance as you grow, by adding storage servers and disks to the system you have.
Recognized by ThinkParQ, the developers of BeeGFS
BeeGFS award · SC24
Bee-Xtra-ordinary
BeeGFS award · 2021
Active Bee
BeeGFS award · SC18
Rising Bee
Why BeeGFS?
As your cluster grows, the file system can become the limiting factor, and expanding it can cost as much as the cluster itself. BeeGFS is the cost-effective way through.
Platinum partner
We’re a Platinum BeeGFS partner in the United States, designing, integrating and supporting BeeGFS on our clusters and on clusters built by others.
The power of clustered storage
Distribute data across any number of storage servers
High streaming throughput and high IOPS
Parallel data access from every compute node
How BeeGFS works
Fast, and built to stay up.
Three services, one file system
Management keeps track of every server, metadata knows where each file’s chunks are, and storage servers hold the data. Once the client knows where to look, it reads and writes directly to storage, so metadata never becomes a bottleneck.
Every target works at once
BeeGFS splits large files into chunks, 1 MB by default and configurable, and spreads them across all of your storage targets, so reads and writes run in parallel. Add targets, and performance grows with them.
Lose a metadata server, keep your files
With buddy mirroring, each file’s record is kept on two metadata servers. If one goes offline, its buddy answers, and the file system keeps working.
Lose a storage target, keep your data
Each chunk is written to both targets in a buddy group. If one target fails, its mirror serves the data, and the file still reads in full.
Maximum scalability
Performance at scale
Flexible striping across storage servers per file or per directory, plus distributed metadata.
Client throughput
High single-stream throughput, and just a few streams can saturate even fast networks.
Linear scaling
Dynamic metadata namespace partitioning keeps growth linear.
Storage pools
Put SSDs and HDDs in different pools, and pin a project to flash for all-flash performance.
Maximum flexibility
Runs where you do
RHEL/Fedora, SLES/OpenSUSE, Debian/Ubuntu and a wide range of kernels, on top of xfs, zfs or other local file systems.
No downtime to grow
Add clients and servers to an existing system without taking it down.
Network failover
Multiple networks with dynamic failover if a connection goes down.
Converged storage
Run client and server on the same machines to turn a compute rack into shared storage.
Case study · AbbVie
1.1 PB of BeeGFS for genomics research.
We deployed a 1.1 PB BeeGFS parallel file system with mirrored metadata servers for AbbVie’s Genomics Research Center, delivering 18 GB/s of throughput to its researchers.
Read the case study →“
In terms of transfer speeds, we’re experiencing three times faster than what we experienced previously.
Martin Forde · Systems Administrator, AbbVie
Built for every performance-oriented environment
HPC
AI and deep learning
Media and entertainment
Life sciences
Oil and gas
FAQ
BeeGFS questions, answered.
Have a question that isn’t here? We’re happy to talk it through.
Ask us →Can you add BeeGFS to a cluster you didn’t build?
Yes. We design and integrate BeeGFS storage for existing clusters, whoever built them, as well as for the clusters we build.
How does support work?
We’re your first line of support. Our engineers handle your questions and issues directly, with escalation to ThinkParQ, the developers of BeeGFS, when needed.
Can we get BeeGFS licenses and enterprise support through you?
Yes. As a Platinum partner, we provide BeeGFS licenses and enterprise support along with your storage, so everything comes from one place.
Can you migrate our data to BeeGFS?
Yes. We can move your data from NFS or another file system to BeeGFS, and plan the cutover around your users.
Storage holding your cluster back?
Engineering the future of HPC and AI
info@advancedclustering.com(866) 802-82223148 Roanoke Road, Kansas City, MO 64111