Draft. This document is awaiting legal review and may change before Galactus Cloud opens to all customers.
This Service Level Agreement ("SLA") is part of the Galactus Cloud Terms of Service. Words defined there have the same meaning here.
During the beta
While Galactus Cloud is in beta, the commitment in section 2 and the service credits in section 3 do not apply, and every Deployment is provided on a best-effort basis. They take effect when we announce general availability, with at least 30 days' notice. We still aim to meet them and to tell you about significant incidents.
1. Which deployments it covers
- Pro sizes (P-…): covered by the commitment below.
- Starter sizes (S-…): no availability commitment. Starter runs on burstable machines for development and small workloads and is provided on a best-effort basis.
2. Commitment
For each Pro Deployment we aim for a Monthly Uptime Percentage of at least 99.5% in each calendar month.
"Monthly Uptime Percentage" is the total minutes in the month, minus Downtime minutes, divided by the total minutes in the month. Months in which the Deployment existed for only part of the month are measured over that part.
"Downtime" is a period of five or more consecutive minutes in which our monitoring cannot open a TLS connection to the Deployment's hostname on port 7687 and complete a Bolt handshake, excluding the periods in section 4. [TO CONFIRM: external monitoring must be in place before this SLA takes effect.]
3. Service credits
If we miss the commitment for a Pro Deployment, you can claim a credit against that Deployment's fee for the month:
| Monthly Uptime Percentage | Credit |
|---|---|
| Below 99.5% but at least 99.0% | 10% |
| Below 99.0% | 25% |
Credits are applied to future invoices for Galactus Cloud, are not paid in cash, are capped at 25% of the Deployment's monthly fee for the month concerned, and are lost if the account is closed. Service credits are your sole and exclusive remedy for unavailability or degraded performance.
4. What does not count as Downtime
- Scheduled maintenance: installing renewed licences, certificate renewals, Galactus DB updates and operating-system patches, normally in the 03:00–05:00 local window for the Deployment's region, up to 4 hours a month. We aim to give 72 hours' notice of maintenance expected to take longer than 5 minutes.
- Unavailability caused by AWS (including failures of an availability zone, region, network or service we depend on), by other third-party providers, by internet or DNS problems outside our network, or by any event beyond our reasonable control.
- Anything you cause or control, including your queries or workloads, running out of disk or memory, your IP allowlist or network, your users and passwords, or changes you make.
- A Deployment that is read-only, stopped or suspended under the Terms (for example for late payment or an Acceptable Use breach), being restored, resized, or being deleted.
- Backups briefly pausing writes, and slower performance that still accepts connections.
- Features, sizes or regions described as preview, beta or test.
5. How to claim
Open a support ticket from your account within 30 days after the end of the month concerned, naming the Deployment and the dates and times of the Downtime. We will review it against our monitoring records, which are final. You cannot claim credits while any payment is overdue.