Your customers don't care about your racks β they care that their services work. The SLA (Service Level Agreement) measures it for each customer, and your reputation sums it up for the whole market.
At any moment, each of a customer's units is in one of these states:
| State | What it means | Billed? | Counts as available? |
|---|---|---|---|
| π’ Running | Placed, reachable, fast enough | β | β |
| π‘ Degraded | Server saturated, network too slow or losing packets | β | β |
| π΄ Disconnected | Server off or failed, or no route to the Internet | β | β |
| βͺ Pending | No server could take it | β | β |
A customer's SLA is its measured availability since the start of its current contract term (signature or last renewal): the share of time its units were running properly.
The SLA is cumulated over the whole term. A customer halfway through its contract absorbs an occasional outage better than one who just signed or renewed β be extra careful with fresh contracts!
You are warned before it's too late: if, at the current pace, a customer would fall below its target soon, its SLA is flagged at risk β a notification, then an alert at the bottom of the screen until things recover.
The bottom bar permanently lists ongoing alerts: failures whose repair hasn't been ordered, saturated servers, missed or at-risk SLAs, an unpaid loan installment, a delivery waiting for roomβ¦
Your reputation goes from 0 to 100 and starts halfway. It slowly follows the quality of service of all your units.
It matters for three things:
Losing a customer because of a missed SLA costs you reputation points on top of the lost income. A good reputation also improves your chances when negotiating a higher price.
Reputation is slow to build and quick to lose. Investing in redundancy, spare capacity and spare parts is investing in reputation.