You can upscale (or scale-up) existing Enterprise (standalone and high-availability) event broker services to a higher service-class level. The reason that you may want to do this is because your existing service requires a higher limit than what you have. The advantage of upscaling an existing service rather than recreating a new service is that you can keep the configuration settings on your existing event broker service, and won't have to re-configure connectivity on your client applications that are connected to your service.
You cannot upscale from standalone services to a high-availability services, or vice-versa. Scaling of Developer services is not supported.
For example, you may require more connections or more queues because your current usage exceeds the limitations for the current service class of your event broker service. For more information about service classes and their limits, see Service Class Options for Event Broker Services.
To upscale your service, follow the steps in Upscaling an Event Broker Service.
Before you upscale your event broker service, ensure that you review the general upscaling restrictions and considerations.
We schedule times to upscale your event broker service as quickly as we can, but you can expect a time of one to two weeks to upscale the event broker service. The more lead time you provide us, the more likely we can schedule your request at the time that you require.
We also recommend that a person with at least the role of Cluster & Mesh Manager Editor opens the support ticket.
If the event broker service to upscale is deployed on your own Customer-Controlled Region (a private network or private datacenter where you control the infrastructure), you must review the considerations and ensure that you fulfill the resource requirements before beginning these steps. For more information, see Limitations and Considerations for Upscaling Event Broker Services
Here are the steps to upscale an event broker service:
- Plan a maintenance window that aligns with your organization and service requirements. For more information, see Plan a Maintenance Window for Upscaling Your Event Broker Service.
- Log in to the PubSub+ Cloud Console and open a support ticket. To do this,
use the Help with My Event Broker Service option and indicate the following information:
- The name of the event broker service in your account that needs to be upscaled. If you are logged into your account, the name is visible in the Service drop-down; otherwise, you can get the name of the service from Cluster Manager
- The new service class you want to upscale the event broker service to.
- The preferred maintenance window times you would like, including the date and time.
We recommend that you identify time periods to upscale your event broker service that work best for you and minimizes the impact to your client applications. During the upscaling time period, the event broker service is unavailable to client applications.
The time to upscale each event broker service is 15 minutes. Though this is the expected time, plan for additional time to align with your organization and service requirements. As an example, you may want to plan for 30 minutes to provide extra time and to align expectations for your organization. The outage of 15 minutes for an event broker service includes:
- the time required to upscale the event broker service on the Kubernetes deployment
- the time to validate the event broker service
In the unlikely situation where the upscale procedure to the service class is not successful, another five minutes may be required to perform a rollback and return your event broker service to its original state. Solace will contact you if rollback is required and collaborate with you to identify the cause.
As the upscaling procedure occurs, there can't be any use of the Message Spool or hard-disks (such as backups) for the event broker service that's being upscaled.
After we receive your request to upscale an event broker service and are ready to accommodate that request, and we have agreed on a time to begin the upscale process, the following occurs (in the following steps, we indicate "Solace" to make it clear it is us who is contacting you):
- Solace sends you a notification containing the date, time, and details of when your event broker service is to be upscaled. You must confirm that date and time, and the request details, and let us know if there are changes.
- Solace sends a second notification to let you know that the upscaling of the event broker service will start during the agreed upon maintenance window.
- During the maintenance window, we upscale your event broker service to the requested service class.
- After the upscale procedure completes (approximately 15 minutes), Solace sends a third notification to let you know the results of the upscale procedure. If the upscale procedure was:
- Successful, proceed to the next step.
- Not successful, Solace contacts you and works with you to correct the problem and to rebook a subsequent follow-up time to upscale the event broker service again.
- After the third notification, client applications can re-connect and start using the event broker service.
After the upscale procedure you can validate that your service has been upscaled. For example, you can check that your service class has been upscaled, or if you were interested in the number of connections increasing, check that the number of connections were upscaled. To do this, go to the PubSub+ Cloud Console and select Cluster Manager in the navigation bar. Click the card for your service and validate the service class or connections in the Configuration tab. For example, after upgrading to the Enterprise 1K (Kilo) service class:
These are the limitations in place for upscaling an existing service class:
- You can only upscale enterprise-grade services (Standalone and High-Availability). Developer services cannot be upscaled.
- You cannot downscale a service (e.g., you cannot downscale an Enterprise 10K service to Enterprise 5K).
- You cannot upscale from a Standalone to High-Availability service (nor scale from a High-Availability to a Standalone service).
- You can upscale directly to any service class level. For example, you could scale from Enterprise 250 directly to Enterprise 100K.
- We only support upscaling event broker services deployed in a Kubernetes environment, and your Kubernetes deployment must support volume expansion. Event broker services deployed using Static Local Provisioned volumes cannot be upscaled. For more information, see Deployment Architecture for Kubernetes.
Before you upscale an existing event broker service, consider whether the new service class you have decided on accounts for future growth?
After your event broker service has been upscaled, all configuration settings for the event broker service remain the same. To take advantage of the new service class limits, you may need to change some of the configuration settings for the event broker service.
The following considerations are only required if your organization deployed the event broker service to Kubernetes as a Customer-Controlled Region (private network (Virtual Private Cloud/Virtual Network (VPC/VNet) or datacenter):
Both the existing and new requirements are necessary until the upscaling of the service occurs. Do you have the necessary infrastructure in place to support the service class you want to move to?
- The Kubernetes cluster where the event broker service runs must support volume expansion. That is, the Storage Class parameter
allowVolumeExpansionmust be set to
true. This setting is required to upscale the event broker service. Once
allowVolumeExpansionhas been set to
true, future upscale requests for event broker services in the same Kubernetes cluster are also possible as long the setting remains set to
allowVolumeExpansion parameter needs to be set on each Kubernetes cluster where an event broker service resides to upscale.
For example, to upscale from an Enterprise 250 to an Enterprise 1K services, you'll require CPU, memory, and disk resources for both the existing Enterprise 250 service and new Enterprise 1K service until the upscaling completes. We recommend that you contact your infrastructure team prior to contacting Solace to upscale your event broker service to ensure that these requirements are met. For more information about the service class requirements, see Resource Requirements for Kubernetes and Default Port Configuration. After the upscaling of the event broker service completes, resources (CPU, memory, hard-drive) from the old service class (that you upscaled from) can be reallocated if they aren't required or you need to optimize your resources in your Kubernetes environment.
We recommend that you choose a size that accommodates not only future requirements (for example, the number of connections), but aligns with your cost structure. If you choose a size that is too large with unused connections, this may not be cost-effective for your organization. Another consideration is that after the service class has been upscaled, you cannot downscale the service class. For more information on choosing the right service class, see Choosing the Right Service Class for Your Event Broker Service.