|Corda Network Foundation||Document history|
Updates and Patching Policy
This policy applies to the Operator for updating software components which have already been deployed onto Corda Network infrastructure.
2 Triggers For Updates
The Operator’s operations team is responsible for monitoring the release of updates to third-party components, including operating systems, middleware and monitoring applications. No stipulation is provided on how to perform this monitoring, although relevant (e.g. vulnerability management) tools may be used for this purpose. Once an applicable update release is identified, the operations team should raise a CR for the update, following the same approval and scheduling workflow as for other CRs.
Updates to Corda components should be requested (via CR) to the relevant software vendor Product Owner following deployment of a confirmed milestone release of that component.
The operations team is also responsible for monitoring the performance of hardware infrastructure and determining where any changes (e.g. to scale up a virtual machine’s cores / memory) are required to deliver optimum performance of the network. Where a hardware change is recommended, the operations team should raise a CR for the update, following the same approval and scheduling workflow as for other CRs.
Updates may also be triggered by the identification of vulnerabilities in Corda, it’s dependencies, or the underlying infrastructure platform. These could come from:
- Vendor or CERT email bulletins.
- Identification of dependency vulnerabilities during CI activities.
- Identification of dependency vulnerabilities during workload or virtual machine image creation.
- Identification of infrastructure vulnerabilities during routine internal or external vulnerability assessments against running infrastructure.
- Incident response activities.
Update activity is subject to the following general principles:
- Immutable infrastructure: To the extent possible, components will be effectively re-deployed rather than subject to updates and changes within the production environment. Re-deployment should be the default practice for all cloud-based infrastructure, with in-situ updates/configuration changes performed by exception only (and for physically-hosted resources).
- Consistency: Except where constrained by specific versioning requirements, updates should be applied to ensure version consistency of all instances of a given component within each environment, and between environments as appropriate.
- Operational continuity: In the context of production environments and test environments with specific agreements in place, software updates should avoid loss of business continuity for users to the maximum extent possible. Planning of the deployment process may seek, for example, to progressively update workers within a clustered service whilst avoiding a situation where all members of the cluster are simultaneously down. This principle also reflects the requirement to retain continuity of user data through the course of the update process (see migration, below).
- Non-automatic: All update operations should, in principle, be consciously instigated by a member of the operations team. Any automatic update features provided by software components should be disabled from the point of installation onwards.
4 Vulnerability (Security Patch) Management
Vulnerabilities will be managed according to the following schedule:
- Critical vulnerabilities (CVSS score > 9.0) will be patched within 7 days.
- High vulnerabilities (CVSS score 7.0 - 8.9) will be patched within 14 days.
- Low to medium vulnerabilities (CVSS score 0.1 - 6.9) will be patched within 30 days.
Where there is information that a vulnerability is being actively exploited “in the wild” (for example, when exploits are available in exploitDB or notification from vendor or vulnerability management software), mitigating action (patching or compensating control) should be taken immediately.
Where a patch is not deployed (or available) within the timescales above, then there must be alternative mitigating action or compensating controls applied.
All updates and patches must follow the same change management process as for new deployments, including the requirements for CR approval, scheduling and testing. See the Change Management Policy for more information.
The operations team are responsible for determining the appropriate method to apply an update in each case. Following the immutable infrastructure principle, component updates should take the form of re-deployment of the relevant resources, with in-situ patching performed by exception only. See the Deployment Policy for more information.
With respect to production environments and test environments with specific continuity agreements in place, all deployment procedures used to perform component updates must put specific measures in place to ensure continuity of operationally significant data (see Backup Policy) before and after deployment. In the majority of cases, is expected that, by design, relevant user data will be persisted in a separate component (database, static file stores etc.) such that frequent functional updates do not automatically require complex data migration.
Prior to designing and/or executing a deployment procedure, the operations team is responsible for confirming all implied data migration requirements in consultation with the software vendor. Where a requirement to migrate data is confirmed to exist, the deployment procedure must document how the migration of data will be performed and validated to remove all risk of irretrievable data loss or disruption of services.
Prior to commencing (or as part of) an update deployment procedure, all operationally significant data associated with the component to be updated should be backed up (see the Backup Policy).