Which profile should a small production site use?

B16 may be appropriate for a small or narrowly bounded workload, but production selection still depends on collection rate, retention, users, reports, integrations and recovery. S32 is the more typical starting point for a moderate mixed production scope. The sizing worksheet and accepted benchmark govern.

Can different hardware profiles be paired in Standard?

Standard ordinarily uses matching nodes so that failover, capacity, spare, firmware and recovery behavior remain predictable. A mixed pair requires an explicitly engineered and accepted exception and cannot assume symmetric surviving-node capacity.

Is local storage capacity the same as retention capacity?

No. Usable retention depends on protection overhead, indexing, compression, event and log volume, rollups, growth, backup space, rebuild behavior and the accepted free-space threshold.

When is 100 GbE required?

It is selected from actual collection, replication, backup, export, aggregation and recovery traffic—not from the profile name alone. High-cardinality cluster telemetry, regional aggregation and large backup or replay windows may justify 100 GbE.

Does the profile include the monitoring software license or managed service?

The hardware profile defines a reference node. Software entitlement, support, implementation, managed operations and hosted federation are separately defined by the order.

LOCAL NX PLANNING PROFILES

PrecisionDCMS NX Hardware Profiles

Reference local-node configurations for Ethernet-native facility, network, compute, storage and cluster monitoring—paired with the workload, availability and benchmark inputs that govern final selection.

Talk to engineering Request an Architecture Review

Access: Open access
Resource type: Hardware profile and sizing guide
Primary audience: Platform engineers, infrastructure architects, procurement, operators and capacity planners
Product scope: PrecisionDCMS NX Lite and NX Standard

REFERENCE, THEN VALIDATE

Hardware profiles create repeatable starting points without pretending every workload is identical.

PrecisionDCMS NX monitors Ethernet-native facility systems, network and OOB infrastructure, servers and BMCs, GPUs and other accelerators, DPUs, Ethernet and InfiniBand fabrics, optical systems, storage, schedulers, orchestrators and shared services.

The B16, S32, A64 and A128 profiles are planning baselines. They standardize a supported hardware shape for common workload bands while preserving the requirement to size active items, value rate, event rate, retention, APIs, reports, collectors, high-cardinality telemetry and surviving-node load for the order.

Final sizing is workload-driven

Final sizing is based on active items, values per second, API cost, event rate, history, users, reports, collector duties, surviving-node load and order-specific benchmark results.

FOUR LOCAL-NODE BASELINES

Select the smallest validated profile that meets normal and degraded-mode load.

Scroll horizontally to see the full table.

PrecisionDCMS NX hardware profile comparison
Profile	Reference compute	Memory	Reference protected data tier	Network	Typical planning use
NX Base — B16	1U, single socket, 16 physical cores / 32 threads	32 GB base; 64 GB recommended for production	2 × 1.92 TB enterprise SSD or NVMe, mirrored	Dual 10/25 GbE; dedicated BMC	Small site, pilot, collector or narrowly bounded local monitoring
NX Standard — S32	2U, dual socket or equivalent, 32 physical cores total	128 GB ECC	4 × 3.84 TB enterprise NVMe, RAID 10 or approved protected tier	Four 10/25 GbE; optional 100 GbE	Moderate production site with mixed facility and infrastructure monitoring
NX Advanced — A64	2U enterprise server, 64 physical cores total	256 GB ECC	8 × 3.84 TB enterprise NVMe, RAID 10 or approved replicated tier	Four 25 GbE; optional dual 100 GbE	Large site, dense infrastructure scope or substantial local retention
NX Advanced — A128	2U enterprise server, 128 physical cores total	512 GB ECC	8 × 7.68 TB enterprise NVMe, RAID 10 or approved replicated tier	Four 25 GbE; dual 100 GbE recommended	Campus or regional aggregation after workload and surviving-node benchmark validation

Reference envelopes, not warranties

These configurations are reference envelopes, not automatic performance warranties. Final CPU generation, memory, storage media, controller, NIC, accelerator, OEM platform and support entitlement are order-specific.

ENTERPRISE SERVICEABILITY

Every accepted NX Node must support controlled operation and recovery.

Compute and memory

Enterprise x86 platform. ECC memory. Hardware sized for the normal workload and accepted surviving-node workload. Virtualization, containerization or bare-metal arrangement defined by the supported baseline. CPU and memory headroom alarmed against the accepted operating envelope.

Storage

Enterprise SSD or NVMe media. Protected boot and application/data tiers. RAID, replication or another approved protection pattern appropriate to the role. Media-health, capacity, latency and rebuild monitoring. Backup and restore separate from local disk protection. Retention sized from ingestion, compression, indexing, rollup and recovery requirements.

Network and management

Dedicated out-of-band management interface. Separate accepted interfaces or logical separation for management, collection, cluster-facing and replication traffic. 10/25 GbE baseline with 100 GbE where workload, aggregation or backup/recovery requires it. Named management identity, certificate and credential control. No direct public exposure of the BMC or local management services.

Power and placement

Redundant power supplies connected to accepted independent sources where the selected server supports them. Rack placement, environmental limits, airflow and service clearances defined in the submittal. UPS or protected power requirement defined by the site architecture. Standard nodes placed in separate accepted rack, power and network failure domains. Console, remote recovery and spare strategy documented.

Supportability

Controlled firmware and software baseline. Vendor support entitlement appropriate to the service. Replaceable-media and hardware-spare plan. Configuration, recovery image and restore procedure. Health, warranty, lifecycle and end-of-support tracking. FAT and SAT evidence for the delivered configuration.

SIZE THE WORKLOAD, NOT THE DEVICE COUNT

Two environments with the same device count may require different profiles.

Collection load

Active normalized monitoring items; sustained and peak new values per second; polling intervals and subscription behavior; API latency and authentication cost; trap, event and webhook rate; log, flow and high-cardinality telemetry volume; number and type of collectors, proxies, agents and exporters; OEM profile transformation and correlation cost.

Data and retention

High-resolution retention; rollup periods; trend, report and query patterns; indexing and cardinality; compression assumptions; backup window and restore objective; replication and surviving-node data responsibility; dataset and export volume.

Users and workflows

Named and concurrent users; dashboards and wallboards; reports and scheduled exports; API clients; tenant and role count; alarm and incident workflow volume; maintenance and change integrations; evidence and audit-query demand.

Failure and growth

Required workload on one surviving Standard node; accepted degradation during replication, rebuild or restore; event-storm behavior; collector backlog and replay; growth horizon and expansion trigger; seasonal, commissioning and test peaks; regional or portfolio aggregation; benchmark safety margin.

AVAILABILITY IS AN ARCHITECTURE

The same hardware profile can be deployed as Lite or Standard.

NX Lite: One NX Node. NX Lite is suitable when the organization accepts a defined single-node interruption and maintains a tested backup, recovery image, restore procedure and hardware repair or spare strategy.
NX Standard: Two matching NX Nodes in independently powered and networked accepted failure domains. The design must validate source distribution, state and data replication, service failover, alarm integrity, identity behavior, backup, restoration and the workload carried by the surviving node.

Standard means two independently serviceable nodes

Redundant PSUs, NICs, CPUs or disks inside one server improve component resilience but do not create NX Standard. Standard means two independently serviceable nodes.

OPERATE THE PROFILE AS A PRODUCT

Placement, configuration and lifecycle controls are part of the hardware selection.

The profile is not complete until the design defines rack and power placement, network zones and interfaces, BMC access, certificates and secrets, time, backup, restore, spare strategy, monitoring of the monitoring platform, vendor support, patch and firmware governance, capacity thresholds and replacement planning.

01
Qualify
Confirm the workload, OEM platform, component compatibility and support entitlement.
02
Configure
Apply the controlled OS, runtime, security, storage, network and monitoring baseline.
03
Validate
Execute profile-specific FAT, performance, failure and recovery tests.
04
Deploy
Verify power, network, time, identity, source access and environmental conditions at site.
05
Operate
Monitor node, storage, network, collection, queue, backup and capacity health.
06
Refresh
Track lifecycle, security updates, compatibility and migration before support limits become operational risks.

The hardware-profile guide includes

Profile comparison and intended use
Common server, storage, network and BMC baseline
NX Lite and NX Standard patterns
Failure-domain placement checklist
Sizing-input worksheet
Normal and surviving-node capacity model
Retention and storage-estimation inputs
Network-interface selection
Backup, restore and spare requirements
Benchmark and acceptance criteria
Procurement and submittal data fields
Lifecycle and expansion triggers

Frequently asked

Questions and answers

Which profile should a small production site use?: B16 may be appropriate for a small or narrowly bounded workload, but production selection still depends on collection rate, retention, users, reports, integrations and recovery. S32 is the more typical starting point for a moderate mixed production scope. The sizing worksheet and accepted benchmark govern.
Can different hardware profiles be paired in Standard?: Standard ordinarily uses matching nodes so that failover, capacity, spare, firmware and recovery behavior remain predictable. A mixed pair requires an explicitly engineered and accepted exception and cannot assume symmetric surviving-node capacity.
Is local storage capacity the same as retention capacity?: No. Usable retention depends on protection overhead, indexing, compression, event and log volume, rollups, growth, backup space, rebuild behavior and the accepted free-space threshold.
When is 100 GbE required?: It is selected from actual collection, replication, backup, export, aggregation and recovery traffic—not from the profile name alone. High-cardinality cluster telemetry, regional aggregation and large backup or replay windows may justify 100 GbE.
Does the profile include the monitoring software license or managed service?: The hardware profile defines a reference node. Software entitlement, support, implementation, managed operations and hosted federation are separately defined by the order.

Related resources

PrecisionDCMS

Field vs NX Selection Guide

Select the architecture from physical-interface, autonomy, availability, data-location, workload and operating-authority requirements.

Read resource Hardware and Configuration

NX-Cloud Service Profiles

Foundation, Standard, Advanced and Private/Sovereign planning envelopes for capacity, retention, isolation, recovery and connectivity.

Read resource PrecisionDCMS

PrecisionDCMS Technical Manual

Controlled product requirements for Field Edition, NX, NX-Cloud, IO, GMS, Site Connector, collectors, software, cybersecurity, commissioning, operations and lifecycle.

Read resource

VALIDATE THE LOCAL WORKLOAD

Move from a reference profile to an accepted node design.

Share the source inventory, active-item and value-rate estimate, retention, user model, availability requirement and deployment constraints. PrecisionX will identify the profile and benchmark path appropriate to the project.

Request an Architecture Review Talk to engineering