126.96.36.199.4 - The repository shall have procedures, commitment and funding to replace hardware when evaluation indicates the need to do so.
- Replacement on a five year cycle or the expiration of a vendor service contract
- Components are routinely swapped out
- Service architecture built in a high availability mode using two techniques 1) Veritas Cluster Server (VCS) 2) Non-VCS with worker instances of things running on separate services monitored by the Apache load balancer
- Typically have 3 to 5 instances running on a server farm but are migrating to virtual machines and using Monit by the end of the year but will continue to run the Apache front-end to send worker processes.
- Triggers for hardware replacement:
- Perform annual capacity planning and also through vendor support/service agreements
- The System Admin checks the logs daily