Upgrading older OCAs in a site, through an RMA or similar process, can provide enhanced reliability and efficiency - often resulting in reduced requirements for power, ports, and rack space. However, when replacing some or all of the OCAs in a site, it is important to understand best practices and plan out the physical upgrade process to ensure minimal negative impact to the traffic offload that the OCAs in a site provide.
The following flow chart describes the general decision process and deployment steps involved when you are replacing the OCAs in a site. Additional detail is provided below the chart.
First, consider the overall space and power resources in the site:
- Do you have sufficient space and power at the site to bring all of the new OCAs online, while keeping the old OCAs online and serving customer traffic?
1 - If YES - you do have sufficient space and power at the site to keep all OCAs (both old and new) online at the same time:
- Install all of the new OCAs.
See this article for more details.
- Wait for all of the new OCAs to complete their burn-in process (typically, a 2-day process) and begin to serve live traffic.
For more information on the burn-in process and how to monitor it in the Partner Portal, see this article.
- Drain the old OCAs of any live traffic, then take them offline and e-waste them.
Your process is complete.
If NO - you do not have sufficient space and power at the site to keep both old and new OCAs online at the same time, consider your backup / failover resources:
- Do you have sufficient spare capacity in one or more backup / failover site(s) to serve the traffic that is currently being served by the set of old OCAs that you are replacing?
2 - If YES - you do have sufficient spare capacity in one or more backup / failover site(s):
- Ensure that you have advertised the same set of BGP prefixes to the the failover / backup site(s) that are advertised to the OCAs in the site where you are performing the replacement.
This step ensures that customer traffic has a graceful failover path as the old OCAs are being taken out of service.
- One at a time, drain each old OCA of any live traffic, power down and uninstall it, then replace it with a new OCA.
- Wait for the new OCA to complete its burn-in process and begin to serve live traffic.
- Repeat steps 2-3 for each old OCA that you are replacing.
- e-waste the old OCAs.
Your process is complete.
3 - If NO - you do not have sufficient capacity in a backup / failover site:
- This case is similar to case 2. However, the lack of backup / failover traffic sources will require you to more carefully and gradually perform steps 2-5 above, pausing after each replacement and monitoring any undesirable traffic shifts to sources outside your network and/or offload degradation.
If you have any questions about this process, please open a ticket.