1. Responsibility regarding the Service
(a)The Company shall guarantee a maximum monthly Failure rate of 1% for the Service during the respective month, and if the monthly Failure rate exceeds the maximum monthly Failure rate, the Company shall issue the customer a Service Credit in accordance with this SLA.
(b)However, if a user uses the Service free of charge under a separate agreement with the Company, the Company shall not guarantee the Service level under this SLA to that user.
Definition
(a) Failure
(1) “Failure” means the case in which the customer is unable to use the Service due to the occurrence of reproducible error when using the Paid Functions of the Service for reasons attributable to the Company. If all requests for external connection fail continuously for more than ten (10) minutes, the period exceeding the foregoing ten (10) minutes shall fall under a Failure of each function
(2) “Paid Function(s) (of the Service)” includes ‘inference with Deployments,’ ‘operation of Deployments’ (acts such as deleting, stopping, or updating a Deployment), and ‘viewing and monitoring Deployments’ etc.
[Functions with No Guaranteed Service Level]
(1) ‘Operating (creating, deleting, etc.) and retrieving Checkpoints’ and ‘Operating Organizations and Projects (inviting the Members to the Organizations and Projects, creating and deleting Projects, etc.)’ are not classified as functions with guaranteed service level because the above functions are not Paid Functions.
(2) ‘A case where a Failure occurs while attempting to create a new Deployment, but that Deployment is not created (used) that month,’ which is one of the functions for ‘Operating Deployments,’ shall not be classified as a Failure because the customer is not paying for that Deployment in that month.
(b) Classification of Failure
(1) Functions are classified into “Failure of Function A”(failures related to inference with Deployment) and“Failure of Function B” (failures other than Failure of Function A). The Company provides more Service Credits in the event of a Failure of Function A compared to a Failure of Function B.
(c) Method of Calculating Failure Rate
(1) Failures are inevitably associated with one particular Deployment. Therefore, Downtime, Monthly Failure Rate, and Monthly Service Fees, defined below, are calculated separately for each Deployment.
(2) “Downtime per Function for Deployment” means the total accumulated time which is recognized as either Failure of Function A or Failure of Function B for the relevant Deployment during the respective month; Provided that, in cases where this SLA is not applicable, the time corresponding to such instances shall not be included as Downtime for either Failure of Function A or Failure of Function B.
(3) “Monthly Failure Rate per Function for Deployment (%)”means the following formula:
100 x ‘Downtime per Function for Deployment’ / ‘Total operating hours in the respective month (e.g., 30 days).’
(4) “Monthly Service Fees (per Deployment)” means the actual amount the customer pays to the Company for the Deployment that experiences Failures in the respective month.
4. Claim for Service Credit and Payment Process
(a) To be eligible for Service Credit, the customer shall submit evidence of the Failure and submit a claim for the issuance of Service Credit to the Company’s customer support center by the last day of the month following the occurrence of the Failure. The claim documents shall include the specific details requested by the Company for processing the Service Credit.
(b) When making payments for the Service Fees, the customer may first deduct any service credit the customer has already received and then proceed to pay for the remaining Service Fees.
5. Cases Where This SLA is Not Applicable
(a)Service credit under this SLA shall not be issued in the following cases:
(1) Where a Failure has occurred to the Service due to factors beyond the Company’s control, such as natural disasters, wars, acts of terrorism, national emergencies, nationwide network failure, or Service disruption due to force majeure equivalent thereto;
(2)Where the Deployment creation and change functions cannot be performed due to allocation failure of cloud virtual machine, etc.;
(3)Where virtual machine used by any replica of the Deployment is in an unhealthy state, without prior notice from the cloud service provider, and inference with Deployment becomes unavailable;(i) Whether a virtual machine is in an unhealthy state shall be determined based on the logs provided by the managed Kubernetes service offered by the cloud service provider such as AKS, EKS and GKE;
(4)Where the reasons are attributable to the customer, such as the customer’s non-compliance with the manual and/or guide for the Service, inexperienced operation, or illegal request;
(5)Where the Service is suspended for inspection according to the prior notice of the Company;
(6)Where the Service is suspended to prevent the spread of accidents that occurred to the customers using the Service;
(7)Where a Failure has occurred to the Service due to an illegal cyberattack from an external source, despite the Company implementing protective measures in compliance with relevant laws and regulations;
(8)Where the Company suspends or terminates the customer’s right to use the Service according to the Terms or a separate contract;
(9)Where a Failure has occurred during the period in which the customer was in arrears for the Service Fees; and/or
(10)Where a Failure has occurred during beta testing, trial use, trial version, preview version, public beta release, etc. (collectively, “Beta Version Services”).
