This article details the recommended RackConnect Operational Limits when building multiple next gen Cloud Servers via the Cloud Servers API.  Note: As performance enhancements are made to next gen Cloud Servers and RackConnect, these results are expected to improve over time. Please keep in mind that all of these results only apply to next gen Cloud Servers.

Please view the RackConnect Network Device Comparison article for details on the performance capabilities associated with each of the Network Devices supported by RackConnect.

LEGEND

Bursting: This term refers to the act of building multiple cloud servers simultaneously via utilization of the Cloud Servers API.  We tested bursting 1, 10, and 30 cloud servers at a time to obtain the results described in this article.

Maximum Number of RackConnected Cloud Servers: The maximum amount of cloud servers you can deploy in your environment before you can expect to begin receiving consistent RackConnect deployment failures.  If you need to deploy more than this amount of cloud servers, we highly recommend contacting your RackConnect Support Team first.  Note: Not all customer environments (Net Devices) will be capable of handling 200 active cloud servers adequately.

API: Optimal number of concurrent builds (bursting): The maximum number of cloud servers you should burst at the same time.  Exceeding this value may lead to a degradation in performance or build errors/failures.

API: Optimal time delay between builds: The minimum amount of time you should wait between cloud server builds.  This applies even between cloud servers built during a single bursting process.  Exceeding this value may lead to a degradation in performance or build errors/failures.

API: Optimal time delay between groups of builds: After bursting a group of cloud servers, the amount of time you should wait before bursting another group of cloud servers.  The "Once All Deployed" time delay signifies that after bursting a group of cloud servers, you should wait until that group is fully deployed with RackConnect before attempting to burst another group of cloud servers.

Total Potential Linux/Windows Builds Per Hour (bursting 30 at a time): The total amount of Linux/Windows cloud servers you can expect to build in an hour by bursting 30 cloud servers, waiting for them to fully deploy with RackConnect, then immediately bursting 30 more cloud servers, repeating this same bursting process for a full hour.

Potential Time to Build 100 Linux/Windows Cloud Servers: The total amount of time you can expect it to take to build 100 Linux/Windows Cloud Servers by bursting 30 cloud servers, waiting for them to fully deploy with RackConnect, then immediately bursting 30 more cloud servers, repeating this same bursting process until you reach 100 cloud servers.

 SUMMARY OF FINDINGS

 These findings indicate:

• If you need to create more than 200 cloud servers in your RackConnect environment, we recommend contacting your RackConnect Support Team first.  Due to the time requirements when deploying cloud servers above 200, it may only be feasible for static environments that will not be changing often.
• For the best results, there should always be a 10 second time delay between API build calls when building multiple cloud servers at once.
• For the best results, you should burst no more than 30 cloud servers at once (with a 10 second delay between builds), wait for them to all go into a RackConnect “Deployed” status, then burst no more than 30 more, repeating the same bursting process until the desired number of cloud servers is reached.
• The "Once All Deployed" time delay signifies that after bursting a group of cloud servers, you should wait until that group is fully deployed with RackConnect before attempting to burst another group of cloud servers.
• As long as the limits detailed above are followed, there should not be a significant time difference, normally less than 20%, between building smaller and larger cloud server flavors.

For further information on utilizing the APIs available, please visit the following links:

Next gen Cloud Servers API documentation

The RackConnect API

How to programmatically determine the RackConnect automation status of you cloud servers