This KMS RFP template gives you a vendor-neutral set of requirements for a key management system (KMS) tender: nine requirement groups, about 90 requirements with priority and the evidence to request, a weighted scoring model, vendor questions, red flags, a proof-of-concept plan and exit clauses. It covers HSM-backed, cloud, external (HYOK, DKE, XKS, EKM, BYOK) and MPC-based key management. Every table can be copied into a spreadsheet.
Table of Contents
- Who this is for and how to use it
- Before you write the RFP
- The KMS requirements checklist
- Weighted scoring model
- Vendor questions to ask in the response
- Red flags in vendor answers
- Proof-of-concept plan
- Contract and exit clauses
- Copy-ready RFP introduction
- FAQ
- References
Who this is for and how to use it
This checklist is for security architects, CISOs, procurement and compliance teams who must issue a request for proposal (RFP), request for information (RFI) or tender for a KMS. It assumes you already know why you need central key management; if not, start with our enterprise key management guide, and use the best KMS buyer's guide to understand the vendor landscape.
How to use it:
- Delete groups that are out of scope and adjust priorities to your risk assessment.
- Keep requirement IDs stable so vendor answers and scores stay traceable.
- Priority values are Must (disqualifying if absent), Should (scored, a gap is acceptable) and Advanced (differentiating, scored as a bonus, never disqualifying).
- Ask vendors to answer each line with: Compliant, Partially compliant or Not compliant, plus the evidence in the last column (a documentation link, certificate number or demonstration).
The requirements are written so that serious products in several categories (HSM-centric platforms, cloud-neutral KMS, MPC-based KMS, key management as a service) can satisfy and evidence them. Items that only some architectures provide are labelled Advanced.
Before you write the RFP
Spend a week on these questions. Most failed KMS projects fail here, not in vendor selection.
Scope.
- Which keys are in scope: data encryption keys, key encryption keys, signing keys, TLS and code-signing keys, secrets-platform master keys?
- Which platforms must call the KMS: Microsoft 365, AWS, Azure, Google Cloud, Salesforce, Snowflake, Oracle, SQL Server, VMware, Kubernetes, custom applications?
- Do any platforms require external key stores (XKS), double key encryption (DKE), customer-managed keys (BYOK) or hold-your-own-key (HYOK)?
Inventory.
- Where do keys live today (cloud KMS, HSMs, SaaS consoles, vaults, scripts)? How many, of which types, with which rotation practice?
- Which key stores must be migrated and which retired?
Stakeholders.
- Security architecture, infrastructure and cloud, application owners, compliance and legal, procurement, and the operations team that will run it. Name an owner for the exit plan.
Constraints and drivers.
- Regulatory drivers, for example DORA (Regulation (EU) 2022/2554) for EU financial entities, NIS2 (Directive (EU) 2022/2555), GDPR Article 32, the Swiss FADP, or sector rules such as FINMA expectations. Record which clause requires which control.
- Data residency and jurisdiction: who may legally compel the vendor, and where do keys and logs sit?
- Certification mandates: FIPS 140-3 level, Common Criteria, national approvals.
Decision criteria. Agree the weights in the scoring model below before you see any vendor response. Changing weights after the demos is how bias enters.
The KMS requirements checklist
Legend: M = Must, S = Should, A = Advanced (differentiating). Evidence is what the buyer should ask for in the response.
1. Key lifecycle management
| ID | Requirement | Priority | Evidence to request |
|---|
| KL-01 | Generate symmetric and asymmetric keys inside the product using a documented random number generator | M | Architecture document; RNG description |
| KL-02 | Explicit key states (for example pre-activation, active, suspended, deactivated, compromised, destroyed) consistent with NIST SP 800-57 | M | State model in documentation; demo |
| KL-03 | Manual key rotation that keeps earlier key versions available for decryption of existing data | M | Documentation; PoC test PT-04 |
| KL-04 | Scheduled (automatic) rotation by key or policy, with configurable intervals | S | Documentation; demo of a schedule |
| KL-05 | Rotation without re-labelling, re-encrypting or re-configuring dependent services where the platform allows | S | Integration documentation; PoC |
| KL-06 | Cryptoperiod enforcement: keys cannot be used for new protection after their period ends | S | Policy configuration screenshot |
| KL-07 | Key destruction with evidence (destruction record retained, metadata kept) | M | Audit log sample |
| KL-08 | Key import (BYOK) with wrapping, and a documented key ceremony | S | Procedure document |
| KL-09 | Key export in standard formats (for example PKCS#8, PKCS#12, KMIP) for keys designated exportable, under permission and audit | S | Documentation; PoC test PT-09 |
| KL-10 | Per-key attributes: owner, usage, tags, expiry, organisational unit | S | Data model |
| KL-11 | Revocation and emergency disable of a key or all keys of an application, taking effect immediately | M | Runbook; PoC test PT-06 |
2. Key custody and protection (HSM, KMS, MPC)
| ID | Requirement | Priority | Evidence to request |
|---|
| KC-01 | Vendor states who can technically use or export a complete key (customer, vendor, cloud provider) | M | Written custody statement |
| KC-02 | Key material is never stored in plaintext at rest; the protection hierarchy is documented | M | Architecture document |
| KC-03 | Support for an HSM backend, with the HSM model and its current FIPS 140-3 (or 140-2 where permitted) certificate number | S | CMVP certificate number |
| KC-04 | The vendor's own module validations are listed per module and version, not per company | M | Certificate list |
| KC-05 | Separation between key store and data store: the party hosting the data does not hold the usable key (HYOK option) | S | Architecture; integration list |
| KC-06 | Customer-controlled hosting option: on premises, customer cloud account or dedicated vendor-hosted tenant | S | Deployment guide |
| KC-07 | Tenant isolation: keys of one business unit or customer are cryptographically and administratively separated | M | Isolation model; test |
| KC-08 | Key never assembled in one place during generation, use and backup (threshold or multi-party computation) | A | Protocol description; independent review |
| KC-09 | Vendor operators cannot access customer key material; technical and procedural controls documented | S | Operations and access policy; SOC 2 or ISO report |
| KC-10 | Multiple key backends selectable per vault (software, HSM, cloud KMS, MPC) from one console | A | Demo |
3. Integrations (cloud EKM and XKS, Microsoft 365, databases, SaaS)
| ID | Requirement | Priority | Evidence to request |
|---|
| IN-01 | KMIP server (state supported version, for example 2.1) with the list of implemented operations | M | Operations coverage table |
| IN-02 | PKCS#11 library (state version) with the list of supported functions | M | Function coverage table |
| IN-03 | AWS KMS External Key Store (XKS) proxy support | S | Documentation; reference customer |
| IN-04 | Microsoft 365 Double Key Encryption (DKE) endpoint and/or Customer Key support | S | Documentation; PoC on your tenant |
| IN-05 | Azure external key management for customer-managed keys (Managed HSM EKM or equivalent) | S | Documentation |
| IN-06 | Google Cloud external key management or client-side encryption support | S | Documentation |
| IN-07 | Database encryption: Oracle TDE, SQL Server EKM, PostgreSQL, MySQL, MongoDB, Cassandra as relevant | S | Per-database guide |
| IN-08 | SaaS customer-managed keys: Salesforce, ServiceNow, Snowflake, Workday as relevant | S | Per-platform guide |
| IN-09 | Infrastructure encryption: disk and VM encryption (LUKS, BitLocker, VMware, Nutanix) over KMIP or PKCS#11 | S | Guide; test |
| IN-10 | Secrets platforms: HashiCorp Vault or OpenBao auto-unseal and seal wrap via PKCS#11 | S | Guide |
| IN-11 | REST API covering all administrative and cryptographic operations, with versioning and published reference | M | API documentation |
| IN-12 | Infrastructure-as-code support (for example a Terraform provider) | S | Provider documentation |
| IN-13 | Client SDKs or an application-facing crypto service for custom applications | S | SDK documentation |
4. Cryptography (algorithms, PQC)
| ID | Requirement | Priority | Evidence to request |
|---|
| CR-01 | AES-128/192/256 with authenticated modes (for example GCM), and key wrap | M | Algorithm list |
| CR-02 | RSA (2048 to 4096 bits) and elliptic curves (P-256, P-384, P-521) for signing and encryption | M | Algorithm list |
| CR-03 | HMAC and hashing with SHA-2 and SHA-3 families | S | Algorithm list |
| CR-04 | Crypto policy: administrators can restrict algorithms, key sizes and permitted operations per vault or key | S | Policy screenshots; refusal test |
| CR-05 | Support for NIST post-quantum standards: ML-KEM (FIPS 203), ML-DSA (FIPS 204) and SLH-DSA (FIPS 205), or a dated roadmap | S | Release notes; roadmap commitment |
| CR-06 | Hybrid (classical plus post-quantum) key establishment or signature modes | A | Documentation |
| CR-07 | Crypto agility: change algorithm by policy without application code changes | A | Demo |
| CR-08 | Cryptographic inventory export (what algorithm protects what), for example a CBOM | A | Export sample |
| CR-09 | Statement of cryptographic module validation status for the modules performing the operations | M | Certificate numbers |
5. Access control and quorum
| ID | Requirement | Priority | Evidence to request |
|---|
| AC-01 | Administrator sign-in through the corporate identity provider (OIDC or SAML) with multi-factor authentication | M | Documentation; demo |
| AC-02 | Fine-grained role-based access control with custom roles scoped to keys, vaults or organisational units | M | Role model |
| AC-03 | Separation of duties: administrators who manage keys cannot use them to decrypt data, and vice versa | M | Role matrix |
| AC-04 | Quorum (M-of-N, four-eyes) approval for sensitive actions: delete, export, restore, policy change | M | Documentation; PoC test PT-05 |
| AC-05 | Runtime authorisation on key use by identity group, not only by individual user | S | Documentation |
| AC-06 | Contextual access conditions on key use (IP range, country, time of day) | A | Policy example; refusal test |
| AC-07 | Explicit deny rules that override allow rules, for incident response | S | Policy evaluation order |
| AC-08 | Several identity providers supported at once (for example Entra ID and another OIDC provider) | S | Configuration guide |
| AC-09 | Optional mutual TLS or client-certificate check on key-use calls | A | Configuration guide |
| AC-10 | Access rules take effect immediately when changed, without restart or redeployment | S | Demo |
| AC-11 | Break-glass procedure with sealed credentials and mandatory review | S | Procedure |
6. Audit and monitoring
| ID | Requirement | Priority | Evidence to request |
|---|
| AM-01 | Every key and administrative operation is logged with identity, time, source and outcome | M | Log schema sample |
| AM-02 | Log export to SIEM (syslog/CEF, Splunk, Microsoft Sentinel, QRadar) in near real time | M | Integration guide |
| AM-03 | Tamper-evident logs (signed or hash-chained records) | A | Technical description |
| AM-04 | Logs record denied operations and policy decisions, not only successes | S | Log sample |
| AM-05 | Configurable retention and an export that auditors can verify | S | Retention settings |
| AM-06 | Health and usage dashboard (requests, errors, latency, key counts) | S | Demo |
| AM-07 | Standard monitoring protocols (SNMP v2c/v3, Prometheus metrics or equivalent) for operations teams | S | Documentation |
| AM-08 | Alerting on key events: expiry, failed approvals, abnormal request volume | S | Alert configuration |
| AM-09 | Compliance reports mapped to ISO/IEC 27001:2022 Annex A.8.24 and your regulatory drivers | S | Sample report |
7. Availability and disaster recovery
| ID | Requirement | Priority | Evidence to request |
|---|
| AV-01 | High-availability design with no single point of failure for key operations | M | Architecture diagram |
| AV-02 | Documented availability SLA for the service (managed) or reference architecture (self-run) | M | SLA text |
| AV-03 | Multi-site or multi-region deployment surviving the loss of a data centre | S | Reference architecture |
| AV-04 | Backup and restore of keys and configuration, with documented procedure and tested recovery time | M | Runbook; PoC test PT-07 |
| AV-05 | Recovery objectives stated (RTO and RPO) | S | Written commitment |
| AV-06 | Behaviour when the KMS is unreachable: documented effect on each integrated platform | M | Per-integration statement |
| AV-07 | Latency figures for typical operations under load, with test conditions | S | Benchmark report |
| AV-08 | Capacity and scaling model (requests per second, number of keys) | S | Sizing guide |
| AV-09 | Rolling upgrades without service interruption | S | Upgrade procedure |
8. Deployment and sovereignty
| ID | Requirement | Priority | Evidence to request |
|---|
| DS-01 | Deployment options offered: vendor-hosted, customer cloud, on premises, hybrid | M | Deployment guide |
| DS-02 | Customer chooses the hosting region or country; no silent relocation | M | Contract clause; documentation |
| DS-03 | Vendor jurisdiction, parent company and subprocessors disclosed | M | Written statement; subprocessor list |
| DS-04 | Statement on exposure to non-EU or non-Swiss legal compulsion and how the architecture limits it | S | Legal and technical statement |
| DS-05 | Runs on supported, non-end-of-life operating systems and runtimes | M | Prerequisites list |
| DS-06 | Air-gapped or restricted-network operation supported where required | S | Deployment guide |
| DS-07 | Key shares or key stores can be placed in several jurisdictions under customer control | A | Architecture |
| DS-08 | Container or Kubernetes deployment supported (for example OpenShift) | S | Deployment guide |
| DS-09 | Automated provisioning of platform wiring (for example identity-provider app registrations) | A | Demo |
9. Vendor and support
| ID | Requirement | Priority | Evidence to request |
|---|
| VS-01 | Vendor operations certified to ISO/IEC 27001 (state scope); SOC 2 report available under NDA | M | Certificate; report |
| VS-02 | 24/7 support with contractual response times for critical incidents | M | Support policy |
| VS-03 | Free or paid proof-of-concept licence, with terms | M | Offer |
| VS-04 | Public product documentation covering installation, administration and integrations | M | Documentation URL |
| VS-05 | Security vulnerability disclosure process and patch timelines | M | Policy |
| VS-06 | Product roadmap shared under NDA with dated commitments for PQC | S | Roadmap |
| VS-07 | Customer references in your sector and size | S | Two references |
| VS-08 | Professional services and training options | S | Offer |
| VS-09 | Escrow or continuity arrangement if the vendor ceases trading | S | Contract text |
| VS-10 | Transparent pricing metric (per key, per operation, per integration, flat) with a worked example for your volumes | M | Price schedule |
Weighted scoring model
Score each group from 0 to 5, multiply by the weight and add. Fix the weights before you read any response. Treat every Must that is not met as disqualifying, whatever the total.
Scoring scale per requirement or group
| Score | Meaning |
|---|
| 0 | Not offered, or no evidence |
| 1 | Roadmap only, or offered with major limitations |
| 2 | Partially met; workaround or custom work needed |
| 3 | Met, evidenced by documentation |
| 4 | Met, evidenced by documentation and demonstrated in the PoC |
| 5 | Exceeds the requirement; verified in the PoC or by a reference |
Example weights (adjust to your risk profile; the sum must be 100)
| Group | Weight |
|---|
| 1. Key lifecycle | 12 |
| 2. Key custody and protection | 18 |
| 3. Integrations | 18 |
| 4. Cryptography | 8 |
| 5. Access control and quorum | 10 |
| 6. Audit and monitoring | 9 |
| 7. Availability and DR | 10 |
| 8. Deployment and sovereignty | 8 |
| 9. Vendor and support | 7 |
| Total | 100 |
Worked example (made-up numbers, for illustration only). Three generic vendors A, B and C, with group scores out of 5.
| Group | Weight | A | B | C |
|---|
| 1. Key lifecycle | 12 | 4 | 3 | 4 |
| 2. Key custody and protection | 18 | 5 | 3 | 4 |
| 3. Integrations | 18 | 3 | 4 | 3 |
| 4. Cryptography | 8 | 3 | 3 | 4 |
| 5. Access control and quorum | 10 | 4 | 3 | 4 |
| 6. Audit and monitoring | 9 | 3 | 4 | 3 |
| 7. Availability and DR | 10 | 4 | 4 | 3 |
| 8. Deployment and sovereignty | 8 | 4 | 3 | 5 |
| 9. Vendor and support | 7 | 3 | 5 | 3 |
| Weighted score (out of 500) | | 376 | 351 | 364 |
| Percentage | | 75% | 70% | 73% |
Calculation for A: 12x4 + 18x5 + 18x3 + 8x3 + 10x4 + 9x3 + 10x4 + 8x4 + 7x3 = 48 + 90 + 54 + 24 + 40 + 27 + 40 + 32 + 21 = 376. Let your spreadsheet do this with a SUMPRODUCT of the weights and the scores. A close result (here, vendors A and C are within a few points) is a signal to run a longer PoC or to re-examine the Advanced items, not to pick on price alone.
Vendor questions to ask in the response
Put these in the RFP as free-text questions; they surface what a checkbox cannot.
- Who, in your organisation or ours, can technically use or export a complete key? Show us the control that prevents it.
- Where is key material generated, stored and backed up, and in which form during each step?
- Which of the requirements marked Must are met by a third-party product, and which are your own?
- Which KMIP operations and PKCS#11 functions are not implemented?
- Which integrations in your list are generally available, and which are beta or planned? Provide the documentation link for each.
- What happens to existing documents, databases or objects when a key rotates, per integrated platform?
- What happens on each integrated platform if the KMS is unavailable for 5 minutes, 1 hour and 24 hours?
- How do you prove the integrity of audit logs to an auditor?
- Which monitoring protocols do you support natively (SNMP, Prometheus, syslog), and which need an add-on?
- How are administrator approvals (quorum) enforced for deletion, export and restore, and can the vendor bypass them?
- What is your support model outside business hours, and what is the contractual response time for a critical incident?
- Which module certificates (CMVP numbers) apply, to which versions, and what is the validation status of the exact version you will deliver?
- How do you support algorithm changes, including ML-KEM and ML-DSA, and what is your dated roadmap?
- What is the exit path: which keys can be exported, in which formats, under which approvals, and which cannot?
- What are the supported operating systems and runtimes, and when does each reach end of life?
- Which jurisdictions can legally compel you or your subprocessors to disclose customer data or key material?
- Who are your subprocessors and where do they operate?
- What is the price for our volumes over five years, including support, HSM or hardware, integrations and growth?
- Describe the last security incident affecting your product, its disclosure timeline and the fix.
- Provide two references with a similar environment and agree to a call.
Red flags in vendor answers
- "We support all clouds" without a per-platform guide or reference customer.
- Custody answers that avoid the question. If you ask who can use a complete key and receive a feature list, assume the vendor can.
- Certificates for the company, not the module. FIPS 140 validation applies to a specific module and version. Check the number on the CMVP site, and note that FIPS 140-2 certificates move to the Historical list after 21 September 2026.
- Rotation that requires a re-labelling or re-encryption project and is not stated as such.
- Authorisation that defaults to open, such as an empty allow-list that admits everyone in the tenant.
- Logs only as local files, or logs that the administrator can edit.
- End-of-life runtimes or operating systems in the supported platform list.
- No exit path: keys cannot be exported in any standard format and the vendor does not describe a migration process.
- Roadmap dressed as capability, particularly for post-quantum algorithms.
- Pricing that grows unpredictably with keys, operations or integrations, and no worked example.
- Refusal of a proof of concept on your workload, or a PoC that is only a vendor-run demonstration.
Proof-of-concept plan
Run the PoC on your own tenants and workloads for two to four weeks, with success criteria agreed in advance. Each test is pass or fail.
| ID | Test | Pass criterion |
|---|
| PT-01 | Deploy to the target environment using the vendor documentation only | Service healthy within the agreed time without vendor shell access |
| PT-02 | Connect two priority integrations (for example Microsoft 365 DKE and a database or AWS XKS) | Encrypt and decrypt succeed end to end |
| PT-03 | Create keys of each required type and algorithm | All created; crypto policy refuses a non-permitted algorithm |
| PT-04 | Rotate a key manually and on a schedule; decrypt data protected before rotation | Old data decrypts; new data uses the new version; no relabelling needed (or effort measured) |
| PT-05 | Trigger a sensitive action (delete, export) as a single administrator | Blocked until the quorum approves; approvals logged |
| PT-06 | Emergency disable of a key or user during active use | Decrypt refused within the agreed time |
| PT-07 | Back up, destroy the primary service and restore | Recovery within the stated RTO; keys usable |
| PT-08 | Fail one node (or site) under load | No failed operations beyond the agreed threshold; failover time recorded |
| PT-09 | Export an exportable key in a standard format and import it elsewhere | Import succeeds; non-exportable key cannot be exported |
| PT-10 | Forward logs to your SIEM and attempt to alter a record | Events arrive; tamper attempt is detectable |
| PT-11 | Measure latency and throughput for your call pattern | Meets the application's budget at peak load |
| PT-12 | Verify administrator sign-in with your identity provider and MFA | Enforced for every administrator |
| PT-13 | Apply a contextual or group-based access rule and test an out-of-policy request | Denied and logged |
Record effort as well as outcome: hours spent, manual steps, and support tickets raised.
Contract and exit clauses
Negotiate these before signature, when you still have leverage.
- Key ownership. Keys and metadata belong to you; the vendor holds no usable copy unless the contract and architecture state otherwise.
- Export and transition assistance. The vendor will export exportable keys in standard formats and support migration for a defined period at stated rates. Agree which keys are non-exportable and what your exit path is for them.
- Data return and deletion. Deletion of tenant data, keys and logs after exit, with written confirmation and a deadline.
- Service levels. Availability target, support response times, service credits and termination rights for repeated breaches.
- Security obligations. Breach notification timeline, audit rights, access to certification reports, and cooperation with your regulators, including DORA ICT third-party provisions where they apply.
- Subprocessors and location. Prior notice and a right to object to new subprocessors or relocation of key storage.
- Jurisdiction and law. Governing law and venue, and the vendor's obligations when a public authority requests data.
- Escrow and continuity. Software or key-recovery escrow if the vendor stops trading, and a minimum support period for the delivered version.
- Price protection. Caps on annual increases and on charges for growth in keys or operations.
Copy-ready RFP introduction
Subject: Request for proposal, enterprise key management system (KMS). [Organisation] is selecting a key management solution to centralise the lifecycle, custody, access control and audit of cryptographic keys across [cloud platforms, SaaS applications, databases and infrastructure]. The solution must keep keys under [Organisation]'s control, support external key management for [Microsoft 365, AWS, Azure, Salesforce, Snowflake, Oracle, SQL Server], expose standard interfaces (KMIP and PKCS#11) and an administrative API, and provide tamper-evident audit logging, quorum approval for sensitive operations, high availability and a documented exit path. Responses must address the requirement groups in Annex A (key lifecycle, key custody and protection, integrations, cryptography, access control and quorum, audit and monitoring, availability and disaster recovery, deployment and sovereignty, vendor and support) using the compliance scale and evidence requested. Shortlisted vendors will take part in a proof of concept on [Organisation]'s environment. Proposals are evaluated using the weighted model in Annex B. Responses are due [date]; questions to [contact] by [date].
FAQ
Q: What should a KMS RFP include?
A KMS RFP should include scope, an inventory of keys and platforms, a requirements list grouped by lifecycle, custody, integrations, cryptography, access control, audit, availability, deployment and vendor support, a weighted scoring model, a proof-of-concept plan and contract and exit clauses. Each requirement needs a priority and the evidence the vendor must supply.
Q: How do I write KMS requirements that are not biased towards one vendor?
Write requirements as outcomes a buyer can test, such as "key rotation keeps old versions available for decryption", not product features. Mark architecture-specific items, such as multi-party computation or a particular HSM, as Advanced and score them as a bonus. Agree the weights before reading any response.
Q: What is the difference between a KMS RFP and an HSM RFP?
A KMS RFP covers lifecycle, policy, integrations and audit across systems. An HSM RFP concentrates on the hardware boundary: certification, performance, form factor and tamper protection. Most projects need both; a KMS RFP should ask which hardware backends are supported and what certificate each holds.
Q: Which KMS integrations should I ask for?
Ask only for platforms you will use in the next 12 months, and require a per-platform guide and a reference customer for each. Typical lists include Microsoft 365 DKE or Customer Key, AWS XKS, Azure external key management, Salesforce, Snowflake, Oracle TDE, SQL Server EKM, VMware, and secrets platforms such as HashiCorp Vault or OpenBao.
Q: How long should a KMS proof of concept last?
Two to four weeks is typical, long enough to test deployment, two integrations, rotation, quorum approval, recovery and failover on your own workloads. Define pass and fail criteria first and measure effort as well as outcome.
Q: Should post-quantum support be a Must in 2026?
For most buyers it is a Should: ask for support of ML-KEM (FIPS 203), ML-DSA (FIPS 204) and SLH-DSA (FIPS 205), or a dated roadmap, and treat hybrid modes and policy-driven crypto agility as Advanced. If your data must stay confidential for many years, raise the priority.
Q: How do I evaluate MPC-based key management against an HSM in an RFP?
Keep the requirement focused on the outcome, for example "no single administrator, node or device can use a complete key" and "FIPS 140-3 validation where mandated", and let each architecture show how it meets them. Compare latency in the PoC for your call pattern. See MPC vs HSM.
Q: Can I get a ready-made version of this checklist?
The tables on this page are designed to be copied into a spreadsheet, with the IDs as the first column. For an assessment of how specific vendors answer these criteria, see our best KMS buyer's guide and the comparison pages.
References