Forum Discussion
SPIisLatency and SPRequestDuration Response Headers in SPO Pages
wbaer, this is primarily for you. I hate to be a persistent pain-in-the-butt ... but I know that I and several other folks out here would love to get some official position on this:
When troubleshooting SPO performance issues, the SPIisLatency and SPRequestDuration headers are critical. For every customer who complains that "SharePoint Online is slow," I can immediately use these two headers to hone-in on whether the issue is page weight, sub-optimal development, potential internal network issues, or some combination of the two.
I wrote a blog post to help users leverage these response headers in troubleshooting (https://sharepointinterface.com/2017/07/07/the-five-minute-page-performance-troubleshooting-guide-for-sharepoint-online/), but for some time now the presence of those two headers on a page that is returned from SPO has been spotty at best. It looked like the headers came back for a while after I sent you and Chris McNulty a note (and spoke with Chris on a couple of occasions), and I officially declared the headers to be back at one point.
At DevConnections last week, I was giving a performance troubleshooting session, went to find the headers, and was once again left holding the bag in front of a live audience. For example, this is from the home page of the SPO root site in one of my tenants:
Notice: no SPRequestDuration and SPIisLatency headers.
Please, please: are these headers supposed to be present? Are they for some reason being phased-out because something "better" is being developed? It's getting harder and harder to tell live audiences, "Hey, they're supposed to be there. Let's pretend they're there." I'd like to know what exactly is going on.
Could you please take a moment to comment?
- Sean
15 Replies
From my blog post (https://sharepointinterface.com/2017/07/07/the-five-minute-page-performance-troubleshooting-guide-for-sharepoint-online/):
UPDATE (3/20/2018): As most of you who have been following-along in your own tenants know, this issue wasn't actually truly resolved last September. For a while, in some cases, it looked like the SPIisLatency and SPRequestDuration headers came back. But the victory was fleeting, and since that time I've continued to get comments from people saying "but I don't see them!" And while I had the headers for a while in my tenant, I haven't seen them in any predictable fashion.
The good news is that after much hounding and making myself a royal pain-in-the-tuckus to Bill Baer and others at Microsoft, it looks like we FINALLY have the right engineering and dev teams engaged to look at this. We got traction on it this week, with multiple repro scenarios and Fiddler traces being passed around ... so I'm truly hopeful we'll see something before long. Stay tuned!
- Yorick Kuijs
Microsoft
Do you know that the IISLatency and RequestDuration values can now be found in the page content? Just search for g_iisLatency and g_Duration.
On modern pages, yes, Yorick Kuijs . I'm moving slowly, but I finally put together a post on how to get the metrics you mention: https://sharepointinterface.com/2019/03/26/obtaining-performance-metrics-for-sharepoint-online-modern-pages/
- wbaer
I'll take that as a thank you ;-)
Please do ... and to back it up, I'll get you a beverage of your choice at SPCNA :-)
(11/5/2017): For anyone interested in the outcome of this issue: Bill and I were at the DEVIntersections conference in Las Vegas this past week. A bunch of us went out for dinner one night, and I took the opportunity to approach him on this topic. That seemed to jump-start things, and he's followed-up with an internal (to Microsoft) email to the developers to get an appropriate response.
When I have some sort of semi-official reply, I'll update this thread. Stay tuned!
Have you received any feedback yet on this issue?
Nothing definitive, no. Bill Baer approached the dev team about it, and they seemed to indicate (or think) that it had something to do with web parts on a page. I called BS on that, because it's happening on blank pages. Bill also didn't seem to buy it, but he said the dev team would need a repro case to make headway. I generated the necessary data to repro the case, and that's where it was left.
I'm planning to send him another note today to see if there's been any progress.
