Supermicro storage servers for Ceph and backup

Two 2U configurations with AMD EPYC: one with 528 TB raw across 24 drives for pure capacity, and another with HDDs, NVMe and 48 cores to store data and run virtual machines on the same node.

Ask for price and lead timeSee the 2 models
Supermicro ASG-2015S-E1CR24L-01-G2ASG-2015S-E1CR24L-01-G2

The 2 models

Supermicro ASG-2015S-E1CR24L-01-G22UASG-2015S-E1CR24L-01-G21x AMD EPYC 9334 · 512 GBSupermicro AS-2015CS-TNR-01-G22UAS-2015CS-TNR-01-G21x AMD EPYC 9455P · 768 GB

Comparison

ModelProcessorMemoryStorageNetworkForm factor
ASG-2015S-E1CR24L-01-G21x AMD EPYC 9334 (32C/2.7GHz)512GB DDR5-48001x 960GB 2.5" SATA SSD + 24x 22TB 3.5" SAS3 HDD1x dual 100GbE QSFP28Rack 2U
AS-2015CS-TNR-01-G21x AMD EPYC 9455P (48C/3.15GHz)768GB DDR5-64002x 240GB 2.5" SATA SSD + 8x 24TB 3.5" SATA HDD + 2x 3.8TB U.2 NVMe1x dual 25GbE SFP28Rack 2U

How much usable space Ceph leaves and how many nodes you need

The datasheet figure is raw capacity, and Ceph spends part of it to survive failures. With 3-way replication it keeps each piece of data on three different nodes and you are left with a third, so the 528 TB of an ASG-2015S-E1CR24L becomes 176 TB usable and the 192 TB of an AS-2015CS-TNR becomes 64. With 4+2 erasure coding each object is split into four data chunks and two parity chunks and you keep two thirds of the space: 352 TB and 128 TB per node. It loses no data even if two nodes fail, but with exactly six nodes, once the second one goes down it stops accepting writes until one comes back.

The node count follows from that. 3-way replication needs at least three, and with three ASG-2015S units you have 1,584 TB raw and 528 usable. 4+2 needs six, because each chunk goes to a different node, and with six of those machines you reach 2,112 TB usable. For both replication and 4+2 we recommend one node above the minimum, so the cluster can rebuild onto the others when one goes down. On top of that you need headroom: Ceph warns at 85% full, so it is wise to plan to stay below 75%.

The choice between the two comes down to the type of data. The ASG-2015S-E1CR24L suits large volumes written once and read rarely, such as backups, archive, video or analytics repositories. The AS-2015CS-TNR carries NVMe for Ceph's internal database and plenty of processor, and it is the one we use for virtual machine disks or for a hyperconverged cluster with Proxmox, where each node stores and runs workloads at the same time. Its two NVMe drives can serve as DB/WAL for the spinning disks or form a small fast pool for virtual machines, but not both with room to spare.

Why buy from us

We deliver the machine installed, and if it ever fails you call the same team that put it into service.

We rack it, configure remote management (BMC) and install RHEL, SUSE, Ubuntu or Proxmox, with whatever runs on top already installed and tested: a Ceph cluster, OpenStack, Kubernetes with the NVIDIA GPU Operator, or Slurm.

When something fails you speak to the people who know that machine because they commissioned it, and if you need it there is 24×7 on-call cover with a contractual SLA. It's the same independent support we provide for IBM Power, Linux and Ceph estates.

If you need different memory, drives or networking, we start from the same base model. Tell us what you need in the custom server configurator and we'll call you back within 24 hours with a proposal.

We are an IBM Business Partner and a Red Hat, SUSE and Canonical partner, which is what counts when the server has to slot into an environment with existing systems and licences.

We work across Europe in all three languages, support included, so if your company has offices in Madrid and Brussels, the same team looks after both.

How we build it

What runs on top of these servers, explained in detail:

Ceph support and consultingCeph for AI and HPC

Support and training for your team

Independent support from the people who installed itCeph trainingCeph Production Engineering

Frequently asked questions

Ceph or a storage array?

Ceph grows by adding nodes, serves block, file and S3-compatible object storage from the same system and does not tie you to an array vendor; in return, it needs at least three servers and someone who knows how to run it. With a few terabytes in a single location, an array is usually enough. We specialise in Ceph and also support clusters we did not build ourselves.

Are they suitable for Proxmox Backup Server?

Yes, and the ASG-2015S-E1CR24L-01-G2 is a particularly good fit: 24 drives behind an HBA for ZFS and two 100 GbE ports so the backup window does not depend on the link. Proxmox Backup Server deduplication means each new backup only takes up what has changed.

What warranty do they have?

Three years parts and labour from Supermicro and, in the first year, advance shipment of the replacement part. On top of that we can provide OS and Ceph support, with 24×7 on-call cover under contract.

Can I mix both models in the same cluster?

Yes. The sensible approach is to separate them with device classes and CRUSH rules (which decide which drives each piece of data ends up on), so backup and archive live on the ASG drives and virtual machine disks on the NVMe nodes. If you mix them in one pool, the slowest drive sets the pace.

Can Ceph run inside Proxmox?

Yes. Proxmox VE has Ceph built in and you manage it from the same interface, so with three or more servers you can run virtualisation and storage in a single cluster. When data volumes or traffic grow we usually split storage nodes from compute nodes, and we decide that with you before ordering anything.

Which one fits what you're building?

Tell us about the workload and the data centre it's going into, and we'll come back with the price, a confirmed lead time and what it would take to get it running.

Ask for price and lead timeWhatsApp+34 91 198 02 43