An Oracle blog about Exadata

  • June 19, 2020

Gen 2 Exadata Cloud@Customer New Features: Scaling OCPUs Without Cloud Connectivity

Manish Shah
Sr. Principal Product Manager

Exadata Cloud Development team is pleased to announce "Scaling OCPUs Without Cloud Connectivity" for Gen 2 Exadata Cloud@Customer

The "Scaling OCPUs without cloud connectivity" functionality allows a customer to change the number of OCPUs used by the guest VM even if the connectivity between Gen 2 ExaC@C and Oracle public cloud control plane is lost. The disconnected scaling operation will allow customers to scale guest VM OCPUs up or down using two special dbaascli commands exclusively designed  to work while there is no connectivity between Gen 2 ExaC@C and Oracle public cloud control plane. The command can be run from any node inside a VM cluster to change the CPU core count for that cluster. If the customer has more than one VM Cluster, they will need to issue a separate command from inside each VM cluster they wish to scale up or down.

The following command can be used to scale the OCPUs up or down in a VM Cluster:

dbaascli cpuscale update --coreCount <core count> --message <message> (example: dbaascli cpuscale update --coreCount 6 --message Scaling_Up_from_4_OCPU_per_VM_to_6_OCPU_per_VM)

The coreCount value refers to the desired value of core count per VM Guest in the cluster. In the above example, the core count in each node will be changed to 6 OCPU per VM. The – message option allows you to associate a message that is visible in the log file located at /var/opt/oracle/log/cpuscale_status folder with the name format cpuscale_status_<timestamp>.log

The second command designed to be used exclusively in disconnected mode is:

dbaascli cpuscale get_status (example: dbaascli cpuscale get_status)

The dbaascli cpuscale get_status command is typically used right after the dbaascli cpuscale update command is issued to get the real time status on the execution state of dbaascli cpuscale update command. It will display various command execution states as it progresses from scheduledrunning and finally to success or failure. The standard linux OS command "top" can issued to find out current OCPUs active in the node (# OCPU = # vCPU / 2) before or after dbaascli cpuscale update command.

The above two commands are designed to not work if issued during the normal connected mode.

Operations and Billing:

Customers can subscribe to be notified when the Infrastructure changes from "Active" to "Disconnected" mode by subscribing to the Exadata Infrastructure - Connectivity Status event. When the customer infrastructure changes status to "Disconnected", Oracle Ops team immediately gets notified and a Sev 2 ticket is automatically opened. Oracle Ops team will first work to identify and rectify the issue if under Oracle control. Otherwise, they will work with the customer to resolve the issue to reestablish the lost connectivity.

During the period of lost connectivity, Oracle will continue to bill the customer for the last value known of active OCPUs to the cloud control plane before the loss of connectivity . The customer billing will transition to the new value (If changed) of configured OCPU when the connectivity is reestablished.

Oracle considers scale up and scale down operations as one of the most important aspects of a true cloud offering and as a result is offering this ability for customers to change their OCPU usage regardless of their connectivity status to the Oracle public cloud. With this capability, customers can continue to scale their critical workloads to meet the needs of their end-users, without worrying about potential side effects of temporary loss of connectivity.

Be the first to comment

Comments ( 0 )
Please enter your name.Please provide a valid email address.Please enter a comment.CAPTCHA challenge response provided was incorrect. Please try again.