I don't know of any templates, and I have not developed any because each environment is so unique they rarely translate (I consult help desk environments). But almost every SLA follows this format:
SLA - clock countdown to an event, usually a boolean evaluation of some condition.
Workshifts - 24/7 shops probably don't need to fuss with this unless holidays aren't supported, i.e. Christmas day. The workshift will allow the SLA clock to pause outside of the predefined time period. Varies by client.
Events - the crux of the SLA. The clock counts down, and checks for something. I.e., after 4 hours check the status of the client's issue. If still open, then do something. If closed, do something else. You can get creative here, from simple notifications to calling elaborate perl scripts.
The SLA is attached to a problem ticket opened by the client. I.e., client loses internet connection. Based on the contract with that client, they have purchased premium service which means we'll call them back within 10 minutes (first event of SLA), and we guarantee they will be back online within 60 minutes (second event of SLA). So an SLA called "premium" is attached to the ticket, and in 10 minutes after opened the 1st event checks to see if issue is acknowledged, then does something. After 60 mins checks to see if still active issue, then does something. What the events do gets into escalation, which is another item...