Search the docs

Hammerspace Hardware and Software Compatibility List

About This Documentation

The Hardware and Software Compatibility List (HSCL) describes the supported and tested hardware and software configurations for a Hammerspace deployment. It covers system requirements, server hardware and CPU compatibility, supported hypervisors, NFS and SMB client compatibility matrices, NAS and object storage compatibility, public cloud support, RDMA and NVMe-oF connectivity, and the qualification process for on-premises server hardware.

This document is intended for infrastructure engineers and procurement teams evaluating or expanding a Hammerspace deployment. Readers are expected to be familiar with Hammerspace software and its capabilities.

For hardware and software questions not covered here, contact support@hammerspace.com with the keyword HSCL in the subject line.

For support resources, see the Hammerspace Support Hub and the Hammerspace Licensing and Delivery Portal.

Do not install additional or third-party software, agents, or packages directly on Hammerspace Anvil (metadata) or DSX (data) nodes. These nodes run a managed appliance software stack; anything installed outside that stack is unsupported and is removed during a software upgrade. Such software can also interfere with node operation. If you need monitoring or integration with an external tool, contact Hammerspace Support for a supported approach.

Prerequisites

It is expected that the reader of this document is familiar with Hammerspace software and its capabilities.

This version of the document is only applicable to Hammerspace version 5.2.

System Requirements

Hammerspace is delivered as an appliance-OS software solution that runs on a wide variety of platforms, including virtual, bare-metal, and cloud environments. Hammerspace includes a base operating system (Linux) as well as a series of enhancements to ensure that the latest and greatest in security and performance are delivered to our customers.

The system requirements are developed from Hammerspace testing and customer deployment feedback. It is not recommended to use less than the specified requirements.

Note that these are bare-metal server minimums. For guidance on virtual machine installation, please see the installation guide.

Table 1. Table 1 Anvil Bare-metal system minimum requirements
Anvil - Bare-metal system, minimum requirements

CPU

1 CPU / 36 total cores minimum3 . Intel Xeon E5 v3 2680 or better.

Memory

128 GB

Network

10 Gbps

Boot drive

512 GB Hardware RAID 1 (mirror)1

Metadata drive

1024 GB NVMe/SSD drive supports a minimum of 100M files2

DSX - Bare-metal system, minimum requirements

CPU

1 CPU / 16 total cores minimum3

Memory

64 GB

Network

10 Gbps

Boot drive

512 GB Hardware RAID 1 (mirror) 1

Data drives

15 GB minimum of any type

Table 2 DSX Bare-metal system minimum requirements

Notes
1) The recommended hardware RAID card is typically what is provided with the server manufacturer. Intel VROC-configured storage is supported for the boot drive only — see the Installation and Licensing Guide’s Intel VROC Support section for requirements (UEFI only, single VROC device set, no additional software mirroring on top). VROC is not supported for the metadata or data drives.

2) Spinning media (HDDs) is not supported as metadata storage, it is simply too slow to be practical.

3) The actual sizing of the Anvil and DSX will depend on the size of the environment

Server Hardware

The following systems have been installed and tested by Hammerspace. This is not a finite list and if your system is not listed here, please see section for on-site qualification to follow the process to work with Hammerspace to qualify your system.

Table 2. Table 3 Server Hardware List
Vendor Server model Supported Notes

Cisco

C220M4

Yes

Cisco

C220M5

Yes

Cisco

C240M4

Yes

Cisco

C240M5

Yes

Dell

PowerEdge C6320

Yes

Dell

PowerEdge R630

Yes

Dell

PowerEdge R730

Yes

Dell

PowerEdge R730xd

Yes

Dell

PowerEdge R930

Yes

Dell

PowerEdge R640

Yes

Dell

PowerEdge R740

Yes

Dell

PowerEdge R740xd

Yes

Dell

PowerEdge R7515

Yes

CPLD 1.07 or later required

Fujitsu

PRIMERGY RX2540 M1

Yes

HPE

ProLiant BL460c Gen8

Yes

HPE

ProLiant DL380p Gen8

Yes

HPE

ProLiant ML310e Gen8

Yes

HPE

ProLiant DL360 Gen9

Yes

HPE

ProLiant DL380 Gen9

Yes

HPE

ProLiant DL580 Gen9

Yes

HPE

ProLiant DL360 Gen10

Yes

HPE

ProLiant DL380 Gen10

Yes

HPE

ProLiant DL580 Gen10

Yes

HPE

ProLiant DL325 Gen10 Plus

Yes

Huawei

RH1288 V3

Yes

Huawei

XH628 V3

Yes

Intel Corporation

S2600WFT

Yes

Lenovo

System x3650 M5

Yes

Lenovo

Think System SR635

Yes

Quanta

D51BP-1U

Yes

Quanta

Freedom

Yes

Supermicro

2029BT-HNC0R

Yes

Supermicro

H12DST-B

Yes

Supermicro

SSG-6028R-E1CR12L

Yes

Supermicro

SYS-1027R-WC1R

Yes

Supermicro

SYS-120C-TN10R

Yes

Supermicro

SYS-220U-TNR

Yes

Supermicro

X10DRi

Yes

CPU Compatibility

CPU Architecture

Currently, Hammerspace nodes are qualified only for Intel and AMD x86-64 CPUs. Non-Hammerspace nodes can be of a different CPU architecture.

CPU Extensions

Hammerspace requires the CPU to support the following feature sets.

  • cx16 (CMPXCHG16B - Double compare-and-swap, introduced with Pentium 4 in 2000)

  • sse4_2 (SSE 4.2, introduced with Intel Nehalem in November 2008 and with AMD in 2011)

NUMA Support

NUMA supports up to two NUMA nodes. If the system is configured with more than two NUMA nodes, it may not install/start properly.

Any power-saving, performance-throttling CPU settings must be disabled on any bare-metal system running Hammerspace software.

Hypervisor

Hammerspace has been tested and is known to work with the following Hypervisors / virtual environments.

Vendor Version(s) Supported Notes

VMware vSphere

6.x, 7.x, 8.x

Yes

Microsoft Hyper-V

2016, 2019

Yes

KVM

Incl. Proxmox

Yes

Field qualified

Nutanix Acropolis

Yes1

Field qualified

XenServer

Yes

Field qualified

Hypervisor compatibility list

1) Hammerspace installs and runs in the virtual environment; however, it may lack platform drivers to show additional information (such as IP addresses) in the native virtualization UI.

Client - NFS

NFS Version 3

Linux

All Linux versions are supported.

Recommended mount options: nolock

UNIX

Solaris, AIX is supported.

Recommended mount options: nolock

NFS Version 4

4.0

Not supported.

4.1

Supported.

The minimum Hammerspace version required is 5.0.

4.2

Hammerspace qualifies Linux client kernels for the NFS 4.2 protocol to ensure all functionality works as expected.

The following kernel versions are qualified to connect using the 4.2 protocol. If a Linux kernel is not on the list, it will automatically be redirected to use version 3. See the Hammerspace 5.2 Administration Guide for further information on how this logic works.

The RHEL/CentOS-family rows below only match a kernel that reports itself as x86_64. The generic "6+" and "7+" kernel rows match any architecture the kernel reports.
Table 3. Linux Client Kernel Support for NFSv4.2
Linux Kernel version(s) NFS protocol version Notes

7+ kernels

Linux 7.x (any architecture)

3, 4.1, 4.2

6+ kernels

Linux 6.x (any architecture)

3, 4.1, 4.2

5.15+ kernels

Linux 5.15.x

3, 4.1, 4.2

(incl. relevant Ubuntu kernels)

5.14+ EL9 kernels

Linux 5.14.x (EL9 builds)

3, 4.1, 4.2

(incl. relevant Rocky kernels)

Debian 5.10+ kernels

Linux 5.10.0-30 through 5.10.0-32 (amd64)

3, 4.1, 4.2

Only build revisions 30-32 are qualified; other 5.10.0 revisions are not.

RHEL 8.10 kernels

Linux 4.18.0-553.x (el8_10, x86_64)

3, 4.1, 4.2

Use 4.18.0-553.116 or later. Earlier 553 builds are missing NFS direct I/O fixes.

RHEL 8.9 kernels

Linux 4.18.0-513.x (el8_9, x86_64)

3, 4.1, 4.2

RHEL 8.8 kernels

Linux 4.18.0-477.x (el8_8, x86_64)

3, 4.1, 4.2

RHEL 8.7 kernels

Linux 4.18.0-425.x (el8_7, x86_64)

3, 4.1, 4.2

RHEL 8.6 kernels

Linux 4.18.0-372.x (el8_6, x86_64)

3, 4.1, 4.2

RHEL 8.5 kernels

Linux 4.18.0-348.x (el8_5, x86_64)

3, 4.1, 4.2

RHEL 8.4 kernels

Linux 4.18.0-305.x (el8_4, x86_64)

3, 4.1, 4.2

RHEL 8.3 kernels

Linux 4.18.0-240.x (el8_3, x86_64)

3, 4.1, 4.2

RHEL 8.2 kernels

Linux 4.18.0-193.x (el8_2, x86_64)

3, 4.1, 4.2

RHEL 7.7+ kernels

Linux 3.10.0-1062, -1127, or -1160.x (el7, x86_64)

3, 4.1, 4.2

Ubuntu 24.04 LTS

3

Use mount bypass option for 4.2

Ubuntu 23.10

3

Use mount bypass option for 4.2

Ubuntu 22.04 LTS

3

Use bypass mount option for 4.2

Ubuntu 20.04 LTS

3

4.2 not tested

Client - SMB

Hammerspace supports several versions of SMB; the actual version used is auto-negotiated between the client and the server.

SMB version Status

SMB version 1

Not supported

SMB version 2.0, 2.1

Supported

SMB version 3.0, 3.0.2

Supported

SMB version 3.1 and later

Supported

Client-Specific Testing (SMB)

Client operating system Notes

Linux (SMB mount)

Not tested but known to work. Best effort support

macOS (all versions)

Supported

Windows (all versions)

Supported

Protocol-Specific Functionality Matrix for SMB

SMB feature Supported Assimilated Notes

Access Based Enumeration (ABE)

Yes

N/A

Requires do-not-cache objective

ACL

Yes

Yes

Share ACL

Yes

No

Security ACLs

Yes

Yes

Alternate Data Streams

No

No

Anonymous (Guest) Access

No

N/A

Auditing

Yes

No

Backup Operator

No

N/A

BranchCache

No

N/A

Change Notify

No

N/A

Continuous Availability

No

N/A

Copy Offload

Yes

N/A

DFS

Yes

N/A

Directory Leasing

No

N/A

Durable File Handles

Yes

N/A

Encryption

Yes

N/A

Hidden/Administrative Shares

Yes

N/A

Homedirs

No

N/A

Multichannel

Yes

N/A

Supported as Technology Preview

Named Pipes

Yes

N/A

Oplocks (batch, level 1, level 2, lease)

Yes

N/A

Persistent File handles

No

N/A

Previous Versions

Yes

N/A

Supported for Share Snapshots, File Undelete, Versioning and Conflict versioning

Remote VSS

No

N/A

RFC 2307

Yes

N/A

Also includes RFC 2307biz

Signing

Yes

N/A

SMB Direct (RDMA)

No

N/A

VSS

No

N/A

Windows Extended Attribute API

No

N/A

Witness Protocol

No

N/A

Storage

The following storage combinations have been tested and are known to work with Hammerspace.

NAS Storage

Most NFS storage exports will work with Hammerspace; however, if the vendor is not listed in this table, support functionality will be covered under type: NFS Other.

Storage Vendor / Model Version Backend RW storage API Integrated Clone & Health Monitoring NFS Assimilation (RW) NFS Assimilation
(RO)
SMB Assimilation

Dell EMC Isilon

7.x, 8.x, 9.2-9.9

Yes

Yes

Yes

Yes

Yes

Dell EMC Unity

N/A

Yes

No

Yes

Yes

No

Dell EMC VNX

Yes

No

Yes

Yes

No

Dell Embedded NAS

Yes

No

Yes

Yes

No

Google Cloud Filestore

N/A

Yes

No

Yes

Yes

No

NetApp 7-mode

8.1,8.2.3

Yes

Yes

Yes(1)

Yes

Yes

NetApp C-mode

8.2, 9.9-9.19.x

Yes

Yes

Yes(1)

Yes

Yes

NetApp FsX

N/A

Yes

No

Yes(1)

Yes

Yes

NetApp Cloud Volume

N/A

Yes

No

Yes(1)

Yes

No

Pure Storage FlashBlade

N/A

Yes

No

Yes

Yes

No

Qumulo

4.3

Yes

No

Yes

Yes

Yes

RozoFS

N/A

Yes

Yes

Yes

Yes

No

SoftNAS Cloud

N/A

Yes

No

Yes

Yes

No

VAST

4.4

Yes

No

Yes

Yes

Yes

WEKA

Yes

No

Yes

Yes

No

Windows File Server

2016, 2019, 2022

Yes

No

Yes

Yes

Yes

Hitachi NAS (HNAS)

Yes

Rclone

Yes

Other NFS

N/A

Yes

No

Yes

Yes

No

1) RW mode is only supported when the backend volume is in UNIX mode.

RDMA

The product supports RDMA using RoCE or InfiniBand. RDMA is supported between the Client and the DSX nodes, as well as between DSX nodes. RDMA is not supported (or used) between the Client and the Anvil (Metadata server).

RoCE:
Support ConnectX-4 and above

The use of PFC, ECN, and DSCP is recommended.

RDMA acceleration of the NFS data path is available only to NFS version 4.2 clients. A client negotiates RDMA through pNFS GETDEVICEINFO: when the data server advertises RDMA and the client is mounted with NFS v4.2, the client uses RPC/RDMA to reach the data server. Clients using any other protocol or NFS version — including NFS v3 and NFS v4.0/4.1 — use TCP for data access and do not benefit from RDMA, even on RDMA-capable hardware.

RDMA and TCP clients can access the same data concurrently; the server provides each client a layout appropriate to its connectivity.

NVMe over Fabrics (NVMe-oF)

The product supports the following NVMe over Fabric storage.

Vendor Version Notes

Hitachi VSP One Block

N/A

Supported with multi-pathing

The Product Object Storage

The following Object Storage vendors and technologies have been tested. For products not listed below, the Generic S3 option can typically be used if the system supports an S3 interface.

Vendor Version Notes

Amazon S3

N/A

Supported Storage Classes: Standard

Azure

N/A

Supported Storage Classes: Hot, Cold, Cool and Archive

Backblaze B2

N/A

Cloudian HyperStore

6.2.x

DataCore Swarm

Dell EMC ECS

3.2.x

Dell EMC PowerScale OneFS S3

N/A

Generic S3

N/A

Includes MinIO, OTC, Ceph

Google Cloud Storage

N/A

Supported Storage Classes: Standard, Nearline, Coldline and Archive

Google Cloud Storage (S3 Interop)

N/A

Hitachi Content Platform

7.3.x

IBM Cloud Object Storage

3.11.x

NetApp ONTAP S3

NetApp StorageGrid

10.4.x

Pure Storage FlashBlade S3

N/A

Quantum ActiveScale

N/A

Scality RING

N/A

Seagate Lyve Cloud

N/A

Snowflake

N/A

Storj

N/A

Wasabi Cloud Storage

N/A

Public Cloud

There are two ways Hammerspace can use the cloud:

  1. Run a Hammerspace cluster (Anvil and DSX) in the cloud (see below for compatible clouds)

  2. Use Cloud/Object Storage from the cloud (see Object Storage)

Hammerspace is supported and available for Amazon AWS, Google Cloud, Microsoft Azure, and Oracle Cloud.

This section records platform availability only. For step-by-step deployment instructions for each platform, see the Cloud deployment guides [Support article may require login to view.] in the Hammerspace Support Hub. For Microsoft Azure, see also Deploying Hammerspace in Azure.

Amazon AWS

Hammerspace is available in the Marketplace.

Google Cloud

Hammerspace is available in the Marketplace.

Microsoft Azure

Hammerspace is available in the Marketplace.

Oracle Cloud

Hammerspace is available in the Oracle Cloud Marketplace, including support for Oracle Dedicated Cloud Region.

On-Site Server Qualification Process

To support the rapid deployment of new systems entering the server market, Hammerspace has developed an on-site qualification process. The process is a part of the standard Proof-Of-Concept (POC) workflow and is fully supported by the Hammerspace engineering team.

It is recommended to involve Hammerspace as the first step in the process. The engineering team can drive the process on behalf of the customer and support remote installs.

Step-by-step workflow:

  1. Verify that the system hardware in question is supported by Red Hat Enterprise Linux 8 with no additional vendor-supplied drivers.
    Link to Red Hat HCL: https://access.redhat.com/ecosystem/search/#/category/Server

  2. Verify that the system hardware options (networking and storage) in question are included as part of the in-kernel tree of drivers for the Linux kernel used by the release you are qualifying against. For 5.2.14 and 5.3.1, that is the CentOS 8 kernel line (4.18.x).

  3. With Steps 1 and 2 executed, proceed to install Hammerspace on the hardware. Hammerspace installation goes through a set of basic tests to ensure that the discovered hardware is functioning properly.

Please contact the Hammerspace team with any questions or feedback regarding the on-site qualification process.