Update on canary - how has our release impact changed?

We had two primary targets for our canary process - our monolith has a main web app and a main api - deployments to those two using normal deploy processes cause 'release impact'.  Namely, all our systems go wonky and the scope of 'what gets disrupted' is sometimes even kinda unknown (or rather, has never been tabulated).
For the backstory, check out the other two posts on this topic:

TL;DR

  1. Customer impact with canary is now 0 (barring a broken deploy)
  2. We can now release at will instead of inside low-throughput release windows
  3. We have fast feedback via metrics that will automatically revert a failed deployment
  4. The actual deploy process is longer, but that's ok - because we don't have a time restriction

Some definitions...

  • Impact window (length of release impact) - how long are users and systems impacted negatively, specifically because of our new code going out (releasing (ok technically deploying))
  • Outage - the application is completely unavailable to serve requests
  • Average latency - measured from IIS responses, using the field 'time_taken', average
    • It's really just for illustration
  • Percentiles are measured from 'time_taken'
    • I chose those three because:
      • 75th being high means things are pretty bad
      • 95th is the one I've seen recommended as 'target to hit'
      • 99th highlights weird horrors

API Release Impact

  • Impact window reduction: 150s -> 30s
  • Outage reduction: 30s -> 0s
  • Average latency during deploy: 8141ms -> 575ms
  • 75th percentile reduction: 13752ms -> 575ms
  • 95th percentile reduction: 19600ms -> 4350ms
  • 99th percentile reduction: 20175ms -> 5900ms

Normal Deploy - API










Canary Deploy - API





Web Release Impact

  • Impact window reduction: 60s -> 0s
  • Outage reduction: 20s -> 0s
  • Average latency during deploy: 4200ms -> 80ms
  • 75th percentile reduction: 7150ms -> 42ms
  • 95th percentile reduction: 14600ms -> 216ms
  • 99th percentile reduction: 20600ms -> 1020ms

Normal Deploy - Web App










Canary Deploy - Web App













Comments

Post a Comment

Popular posts from this blog

Learning through failure - a keyboard creation journey

Image
tl;dr Built a custom keyboard cuz I have weirdo nerve pain issues.   Learned a lot cuz I failed a lot.   Maybe 'do it right the first time' is a bad frame of mind.   The keyboard ended up being kinda nifty.   It has lights, and a trackball, and QMK, and it helps my issues.   \o/ ( tl;dr is internet lingo that means, too long; didn't read ) Pain as a key ingredient An old proverb says, 'necessity is the mother of invention', and that is how this story begins. Over the last few years, age or stress or something has caught up with me and I developed the need to change my human-to-computer interface.  Mousing changed to trackballing.  The traditional keyboard was swapped out for an ergonomic, and then split keyboard.  Complicated multi-finger-and-hand key combinations were replaced by single keypress macros.  And yet more was still needed. It's often said that change only happens with pain - indeed, the organizations many...
Read more

Learning Opportunities - Watching/listening list

Over the last year I've spent a lot of time learning (thank you, long commute & lunch hours).  Many topics ranging from Chef to DevOps to architecture concepts to management concepts and everything in between.  Why have I done this? DevOps Demystified with Ben Rockwood: https://www.youtube.com/watch?v=h5E--QSBVBY DevOps at ancestry.com: https://www.youtube.com/watch?v=pF5ya0q53kk Those two presentations were key in sending me down this road.  First, and most important, was Mr. Rockwood's explanation as to the real 'why', and the point behind DevOps from an operational perspective.  Second, and no less important but for a different reason, the continuous delivery system at ancestry.com really made it real to me because they are a 'Windows shop'.  So much of what I'd been reading about Chef/config mgmt. relied on amazing tools (see the error there?) only available to the Linux world.  Suddenly here's this guy telling me that it's possible...
Read more

DFSR - eventid 4312 - replication just won't work

This warning isn't documented that well on the googles, so here's some google fodder: You are trying to set up replication for a DFS folder (no existing replication) Source server is 2008R2, 'branch office' server is 2012R2 (I'm moving all our infra to 2012R2) You have no issues getting replication configured You see the DFSR folders get created on the other end, but nothing stages Finally you get EventID 4312: The DFS Replication service failed to get folder information when walking the file system on a journal wrap or loss recovery due to repeated sharing violations encountered on a folder. The service cannot replicate the folder and files in that folder until the sharing violation is resolved.    Additional Information:  Folder: F:\Users$\user.name\Desktop\Random Folder Name\  Replicated Folder Root: F:\Users$  File ID: {00000000-0000-0000-0000-000000000000}-v0  Replicated Folder Name: Users  Replicated Folder ID: 33F0449D...
Read more