SLA for Luminis
We are planning on going live with Luminis IV this Fall with parallel deployment. One of the topics that has come up recently is the concept of an SLA for Luminis. I’m wondering if any of you have created a document or policy that you could share which outlines the level of service you provide for the Luminis portal.
What sort of expectations do your users have for the portal?
Do you provide 24x7 uptime?
Do you have scheduled maintenance time (e.g., for backups or hardware maintenance)?
How often do you have unplanned downtime?
What else should we consider?
sla
I'm not sure if it is written out specifically as a policy or agreement, but our rules concerning Luminis (and all public servers) are:
Students use our portal for classwork, registration, bill paying, parking tickets, etc.. basically a one stop shop. So it is expected to be a reliable service.
24/7 in general,
2 week notice on outages which are usually scheduled for low-use periods like late Friday night or an early morning,
No unplanned downtime.
Your SLA is going to be pretty dependent on how much functionality your Portal has, and whether those functions can be accessed outside of the portal. For instance, it is not possible for one of our students to get to self-service without going through our portal.
That said, it is pretty hard to guarantee 24/7 considering that the Database and LDAP are single points of failure. I've heard that some schools have managed to cluster the resource tier, and I assume some schools are using clustering on the DB. For us, the reliability of the ldap and db was good enough without resorting to clustering.
sla / clustering
Thanks for the info.
We considered clustering the DB, but we hadn't heard of anyone clustering the resource tier. Plus, we've heard that Luminis V will support clustering at all tiers. So for now, we've decided to stick with a single LDAP and DB server.
I'm assuming you are able to do backups without any downtime. We have not yet dug into the details here, but I'm hoping this is the case.
Also, during development, we have found that almost any change requires a Luminis stop and restart -- publishing a new channel, changing a timeout value, etc. I'm hoping that most of these things only end up requiring a web tier restart, and that we can use load balancing scripting to automatically take the web server out of the pool, restart, then add it back in (and then repeat with the other web server). But if any routine maintenance work requires a restart of the resource tier, then we'll have to build in some planned downtime into our SLA.