Skip to contact
Official IBM training · HPC and high performance computing

HPC training
for enterprise clusters

Official training in IBM Spectrum LSF (workload scheduling), IBM Storage Scale (GPFS) and Elastic Storage Server. Five official IBM courses for teams running HPC clusters in production — from batch scheduling to the parallel filesystem. Enterprise alternative to Ceph and SLURM when the business needs vendor support and SLAs.

See course catalog Request a quote
Official partner IBM Business PartnerCertified official training
[ 01 ]Two admin layers you can't split

An HPC cluster is two administrations held together with tape.

A high performance computing cluster has two technical layers that usually get learned separately and administered together. On one side, workload scheduling (LSF, SLURM): decides what runs, when, with what priority and against which SLA. On the other, the parallel filesystem (Storage Scale/GPFS, Ceph, Lustre): serves data to hundreds of nodes simultaneously without collapsing. This hub groups both because in production no one keeps them running in isolation for long.

Official IBM training in Spectrum LSF (H010G), Storage Scale / GPFS (H005G, H006G, H008G) and Elastic Storage Server monitoring. Delivered directly by our own deployment engineers, not outsourced instructors.

If you need other pieces of the stack: Ceph SDS for open-source distributed storage, advanced Linux as system base, our IBM Storage solutions for enterprise backup and archival, and IBM Spectrum Compute · HPC consulting when the cluster needs to be built, not just learned.

CATALOG5 courses // FAMILIESLSF · Storage Scale · Elastic Storage // ALT.SLURM · PBS // LANGUAGESEN · ES · FR // MODEonline · in-company
admin@hpc-master · ~ · bash
# submit a GPU job to the "gpu" queue reserving A100s in exclusive mode
$ bsub -q gpu -n 8 -gpu "num=2:mode=exclusive_process" \
    -R "select[gpu_model0=NVIDIAA100]" ./train.sh
Job <18274> is submitted to queue <gpu>.

# check job status with useful columns
$ bjobs -o "jobid stat queue exec_host time_left"
JOBID  STAT  QUEUE  EXEC_HOST       TIME_LEFT
18274  RUN   gpu    2*gpu-node-04   3:47:12

# inspect a queue: priority, pending/running jobs, limits
$ bqueues gpu

What you'll be doing from day one of the H010G course.

[ 03 ]Frequently asked questions

Common questions about HPC

What is HPC (High Performance Computing) and what is it used for?

HPC (High Performance Computing) is the discipline of pooling tens or hundreds of servers into a cluster to run workloads a single server can't handle: numerical simulation (CFD, weather, chemistry), large-scale AI model training, financial risk analysis, seismic processing, semiconductor EDA or rendering. An enterprise HPC cluster needs at least two layers: a workload scheduler that decides which job runs where and when (LSF, SLURM, PBS), and a parallel filesystem that serves data simultaneously to all nodes (Storage Scale / GPFS, Lustre, Ceph).

How does IBM Spectrum LSF differ from SLURM?

SLURM is a free scheduler widely used in academic HPC. IBM Spectrum LSF is the enterprise alternative: it covers contractual SLAs, complex cross-project prioritisation, fine-grained management of expensive software licences, advanced fairshare policies and official IBM support. When the business needs execution-time commitments and auditable trace, LSF fits better. When the priority is zero cost and academic flexibility, SLURM is the way. On the H010G course we run a practical comparison on the cluster.

Do I need Storage Scale (GPFS) if I already have LSF?

Not strictly. LSF can run on NFS or any shared filesystem — it's storage-agnostic. But in serious HPC, NFS runs out of headroom as load grows and becomes the bottleneck. Storage Scale (GPFS) is a parallel filesystem designed so hundreds of nodes can access the same data without degradation. You can also swap in Ceph (open-source) or Lustre. The LSF + Storage Scale combo is IBM's reference enterprise HPC stack.

Can Ceph replace Storage Scale (GPFS)?

Depends on the workload profile. Ceph is powerful and flexible for distributed object, block and file storage — zero licence cost and full control. But on HPC workloads with lots of small I/O and latency-critical paths, Storage Scale (GPFS) still performs better thanks to years of specific tuning. Ceph wins when flexibility, multi-tenancy and mixed workloads matter most. At SIXE we train and deploy both — the decision depends on the actual cluster profile.

Can the training be delivered in-company on our real cluster?

Yes. Every course in the HPC hub can be delivered in-company, adapting the syllabus to your LSF and Storage Scale version, your network topology and the applications actually running in production. We work against your pre-production environment and you leave the course with a configuration effectively ready to promote. Request a quote with number of attendees, versions and location.

Are the courses online, in-person or hybrid?

Both, in three languages (EN, ES, FR). Live online with virtual classroom and a real dedicated cluster per attendee, or in-person in-company at the customer site. Hands-on quality is equivalent — labs run on the same environment.

Does SIXE also deploy the cluster, or only train?

Both, as separate services. Beyond the training catalog, SIXE also offers consulting, deployment and integration of IBM Spectrum Computing HPC clusters (LSF + Storage Scale + Elastic Storage Server). Many customers hire the deployment first and then train their internal team to operate it.
[ 04 ]Beyond training

We also deploy and consolidate the cluster

Solution · IBM Spectrum Compute HPC

Consulting, deployment and integration of HPC clusters

Training the team doesn't replace deployment. Someone still has to size the network, decide the storage architecture, integrate with existing systems and — if you're coming from SLURM or PBS — run the migration without stopping production. That side we handle separately, as consulting.

See full solution →
Alternative · Ceph SDS

Ceph for distributed storage

Open-source alternative to Storage Scale for HPC clusters that prioritise flexibility and cost over vendor support. Our own training and consulting.

See Ceph hub →
[ 05 ]   Request information

Tell us about your cluster

Tell us how many you are, which scheduler and filesystem you run (or which scheduler you're coming from) and whether you prefer online or in-person. We come back with concrete format, dates and instructor team.

Contact us WhatsApp See full catalog
Phone+34 91 198 02 43
PriceOn request per group
LanguagesEN · ES · FR
SIXE