Last January I shared some thoughts on keeping hosted services up and running 99.9% of the time. eMarketing Websites and portals often do not have the luxury of going down for 3 hours of maintenance (or even 3 minutes), leaving me to often envy the planned maintenance notices I receive from Salesforce when logging in
.
As you can see from the daily report below, our monitoring service loves to occasionally tease us with uptime reports hovering at 99.89% uptime.
Some key things we observe and monitor:
- "Downtime" in the report is measured in blocks of 5 minutes, so what may have actually been a 10 second outage gets reported as a 5 minute outage, skewing the actual results.
- Sub-second response time (< 1000 milliseconds), while not contractually in the SLA, is a key metric in determining site performance and customer experience.
- Customers who deploy changes to a staging server first (available in Unlimited Edition) have a much higher likelihood of sustaining 100% uptime day over day.
- Reliance on a good datacenter partner is critical. Our offices are not the most luxurious, but we spare no expense on our datacenter location and services.