Navigation

Get in touch
Logo
News

HSM architecture

The right HSM. Before you invest.

An HSM must handle your real key operations and fit your applications, security requirements and operating structure. We translate these requirements into a well-founded device and architecture decision, before hardware is procured or you commit to a cloud service.

man and woman sitting at table — illustrative image
Selection matrix with rationale · Planning and implementation by OTOKO®

Your brief for OTOKO®

What we take care of for you.

A figure for signatures per second says little as long as the algorithm, key size, parallel sessions and network latency are unknown. We record, for example, certificate issuance, login, document signing or data decryption separately. Peak load, retries and behavior in the event of a node failure belong in the load profile. The required APIs, operating systems and client libraries also help determine which platform makes sense in your environment.

The possible scope of services

  • Record use cases, key types and load profile
  • Compare devices and services by interfaces, security requirements and operation
  • Plan partitioning, site failure and backup
  • Check evidence for specific hardware, firmware and operating modes
  • Assess integration risks with a scoped test

We define the specific scope, your involvement and the acceptance criteria before the start.

Technology explained clearly

How we carry out the task.

01

Mapping security boundaries and failure behavior

The architecture separates applications, administration, backup and key custody. A partition is a logical separation, but it does not replace every organizational or physical separation. We clarify which keys may be replicated, who can extend a cluster and what dependencies exist between sites. If the HSM fails, an application must not silently fall back to unprotected key files. Certificate status and Security Policy are checked for the specific module; a product name or an algorithm validation is not enough.

02

Validating the selection with a representative test

Before final approval, we test typical operations with the intended client connection. Throughput, latency and error handling are measured in the agreed profile. The result includes assumptions about growth, licenses and operation. To get started, we need an overview of applications, existing key types, site requirements and the evidence requirements your organization actually has to meet.

03

Network device, PCIe card or managed service?

A network HSM can provide a central cryptographic function to several applications. In that case, the network path, authentication and tenant separation become part of the architecture. A PCIe card ties the function more closely to the host; a second machine needs its own availability concept. For a managed service, we review the available APIs and how administration is divided. We do not base this decision on the purchase price alone: operating effort, reachable sites, maintenance windows and a later switch all belong in the comparison.

04

Turning a performance specification into an acceptance test

For the pilot, we describe a complete business transaction: which call reaches the HSM, how many calls does each transaction generate and when is a response too late? Alongside average values, we record high latency percentiles and behavior under peak load. A test with a single signature does not reflect parallel clients, connection setup or a node failure. You receive the measured conditions and the remaining headroom, so that procurement is based on a traceable load profile.

05

Example: building a central signing platform

Several applications are to sign through a shared infrastructure in the future. OTOKO® assigns keys and responsibilities to the applications, checks the required mechanisms and compares a shared platform with separate instances. In the pilot, we test not only successful signatures but also missing permissions, exhausted session resources and the shutdown of a node. The result is a well-founded architecture decision with integration effort, license needs and open dependencies, not a blanket recommendation for the largest device.

Meeting room at the OTOKO® Cologne office

A verifiable result

What you keep working with.

  1. Selection matrix with rationale
  2. Target architecture with security boundaries
  3. Test plan and open procurement questions

The handover brings together implementation and documentation. Together, we review the agreed cases and record any remaining tasks.

Before the first step

Your questions about HSM architecture.

Is the most expensive HSM automatically the best choice?

No. Suitable interfaces, security boundaries and a robust operating model are decisive. Capacity or functions you do not need can increase cost and complexity.

Does a FIPS validation carry over to every firmware version?

No. We check the certificate, Security Policy and approved configuration. New firmware or a different operating mode can require a separate assessment.

Which documents speed up the selection?

Helpful inputs are a list of applications and interfaces, existing HSM and client versions, the algorithms used and expected call volumes. Add requirements for sites, downtime and evidence. We can determine missing figures together during the assessment.

Can a logical partition replace a dedicated device?

For some separation requirements, it can be enough. However, we check which resources, administration and failure causes remain shared. We do not treat a logical separation as equivalent to a physical or organizational separation without checking.

How do you take total cost into account?

We look at procurement or rental costs, options and licenses, redundant capacity, backup, client integration and ongoing operation. This makes it possible to distinguish a low entry cost from a solution that is viable over the planned period of use.

Is a proof of concept required before procurement?

If application compatibility is unknown, the load is demanding or a migration is involved, a scoped pilot makes sense. For a documented standard integration, a targeted compatibility check can be enough. The decision depends on the specific risk.

Your project

Which task would you like to solve?

Describe your situation and the desired result. The selected service will be included in the contact request.

Request this service

Our Partners

  • Microsoft
  • Microsoft Azure
  • Amazon AWS
  • Google Cloud
  • Thales Group
  • Arrow ECS
  • Vodafone
  • IBM
  • Veeam
  • Atlassian
  • JetBrains
  • NinjaOne
  • OPSWAT
  • Utimaco
  • Eviden

Accessibility

Adjust the display to suit your needs.

A simple version is not available for this page yet.

Settings currently apply to this visit. Allow saving in Cookie settings to remember them.