I can clear my browser cache, and my personal summary page takes 5 seconds to load from the moment I hit enter. Details pages vary, depending on what resources are shown (we're using Node Details by machine type), but they are anywhere from 5 seconds for a 3750 to about 10 seconds for a fully loaded Nexus 7k. Of course these times can vary depending on the over-all load of Orion at any given moment.
When I first took over our Solarwinds, pretty much all the page load times were slow, but Node Details was the worse. The more I learned about Solarwinds, the more I found that mistakes that had been made with the implementation.
For example, the database had been placed on a SAN tier 2 LUN. I know that Solarwinds generally recommends a local RAID, but my research had turned up info that a good SAN can actually be faster than local RAID. Once we got our database moved to a Tier 1 LUN, that made a noticeable improvement.
The node details page load times would consistently be well over a minute. Sometimes, over 2 minutes. HUGE improvements were made with the Node Details by using View by Device Type. Originally, they were using the same general Node Details for every platform. So, for example, a 3750x Node Details would still have the resources for a Nexus chassis, or a UPS.....and even though there is no data and the resources were not necessarily shown on the page, it definitely had an impact on page load times.
Another item that still impacts our performance is that all network syslog is being sent to Orion. Well, in troubleshooting all the performance issues we had, I found out that Orion was never intended for this. For a medium/large enterprise, you should really be sending logs to a tool like Kiwi or Splunk. Then you can forward what you need in Orion from those tools. Our volume of syslog has a huge impact on our database performance. SNMP traps are a close second. Once we correct this, I expect performance to improve. I also plan to lower our polling intervals once that is done.