lawful evil with neutral tendencies • fixated on deterministic outcomes • probably not a shapeshifting, mind- controlling lizardman • InLinked not BookFace • @mykelalvis • blog.automationmonkey.net • works in a regulated industry barely verbal monkey
or ops or analysts or testers • write code or deploy code or other stuff • maybe even document that stuff • probably not • might realize how hard your stuff is to deal with • well, probably not buncha little crybabies...
process • all process has a flow • all flows change over time • jobs are executions of these flows that we get paid to execute • we use process to increase more of the -ilities the pedantic view from 30,000 feet
platform • release ◦ production of an immutable artifact • version ◦ signature of a( generally released) api ◦ that’s very important to my story • deployment ◦ make an artifact available in an environment(see what I did there?) mr webster is rolling over
the belief that all events are caused by things that happened before them and that people have no real ability to make choices or control what happens • a theory or doctrine that acts of the will, occurrences in nature, or social or psychological phenomena are causally determined by preceding events or natural laws From http://www.merriam-webster.com/dictionary/determinism still rolling over
is prolly error-prone • it has way more steps that it should • it is difficult to replicate • it is human-centric • delivering your stuff to the customer is literally your only job and frequently it’s not done well so?
• i only want to be disturbed if something is actually wrong • automated deployment • automatic maintenance of the deploy • documentation back to source • reliability • repeatability • auditability the goal
what I’m doing) • definitely works for you • definitely challenging to maintain • again, marriage • this time your spouse is a monster perversions of science
immutable delivery allows ignoring change requests for existing things • immutable delivery is intrinsically repeatable (by my definition) • any given system must be stable • so make it stable and if it needs to change, replace it • but what to replace it with?
deploys • deployment of an artifact is versioned • versioned deployments are trackable • (remember that versions imply releases) • even versioned deployments are arbitrarily complex • how do we address the complexity of the deployment itself?
• encapsulated complexity must be versioned • versioned complexity must be orchestrated • and tracked • and audited • and blah blah blah • in short, and to use a dirty word in the DevOps space, enforcement
an asshole • unless it does • you must decide what you things you want to be deterministic • the costs of determinism can be quite high • you need to get the process internalized within the executors
orchestration • bpmn 2.0 is a dumb, old technology • engines running it are typically rock-solid • they can be kinda clunky to deal with • but they’re like democracy • if you have cable from a specific large cable provider and your guide button works, you’re welcome • all 25 million of you
bit clunky • versioning wasn’t impossible, just hard • automated deployment was very hard • the results weren’t very deterministic • so there needed to be a better way • i won’t bore you with all the requirements, but there are several
could do it better ◦ i think i did • backed by a few very dumb technologies ◦ because your neomania is pointless • and a couple of smart ones ◦ because even old dogs learn new tricks
one was less extensive • (i was very, very wrong) • a packaging method for data • produces released artifacts of datasets • provides key-signed provenance of the data • produces what amounts to a zip file with some very important metadata
functionality • produce out-of-band data management • handle provenance of data • provide validation of data integrity what have you done for me lately?
pictures on the internet and preemie-baby incubator management • we work much more like the latter than the former • if you put cat pictures on the internet and someone pays you to do that, congrats no, seriously...why bother?
$500M in 2015, on our way to $1B • brand-promise is everything • healthcare data is both complex and complicated • CleanFuel helps us decrease the complexity (it can’t generally help with the lack of simplicity) all about the benjamins
kentarofujii
occhi
ykanoh
ymotongpoo
persolserverworks
kaminashi
sudoakiy
y0hgi
yuuuuuuu168
yu23ki14
bitkey
shogo4131
427marketing
destraynor
inesmontani
leimatthew05
brad_frost
caitiem20
tamaranovitovic
ammeep
akademiya2063
katiecoart
jonoalderson
![@mykelalvis • [email protected] • devops coach at Cotiviti Labs •](https://files.speakerdeck.com/presentations/c831064e87544eaba69856af07d0ca7a/slide_1.jpg){kind=link}
