Consider that halfway through a day you do a release, and that release slows down your site by 25%, or equally speeds it up. It's good practice and of course everything should go through NFT and be spotted beforehand etc BUT sometimes things happen, and when you look back on results you need to add context to them. Maybe you have an incident and the site is down for 1 hour - but then you do a weekly report, hand it off to someone who didn't know about that downtime - you end up drawing on the thing with MS Paint and that's a pain...
What I am thinking of is a 'Platform Events' function, maybe as an extension to 'Maintenance'.
I want to be able to select a point in time, or a range of time, and assign to that point of time a simple event like 'Sprint 7 Deployment' that required downtime.
It would be awesome if I could even mark for this time period to be excluded from SLA, or excluded from SLA2 (then in the report you could have true SLA and SLA2 representing SLA against expected uptime - hence it shows unplanned downtime)
Similarly then when looking at trend graphs there would be great visual cues for possible performance changes. I mean, next week I'm making a range of small Akamai changes - I want to be able to mark that on and 'tell a story' with the data that you let us gather which again, in the eyes of the greater business shows future benefit.
To be honest if that is something that you could put into place I would love you to expose an API - maybe a simple webhook even for driving these events. When we have Automated Deploy fully up I want it to ping into your platform and create those events for me. When I get a huge traffic spike on Akamai I want to get it to email PagerDuty, and PD can hit your webhook.
It's all about context with me and the greater context you can give the greater value can be delivered. I hope this suggestion makes sense and you can let me know what you think!
N.B. Here is a really basic high level view of what I mean by building context. Etsy are great, and so is their devops blog - but I do not have the resource to build dashes from scratch like they do. Check out P88, they're consuming 3 provider's CDNs but they are also throwing on the times of their deploys. I mean, those guys deploy tens of times a day, but you see my point?
Update: Adding Events via the UI is now implemented since release 7.8 (22 oct) I can unfortunately give no updates regarding Analyze Metrics/Trends as of now.