devops

There is so much more work needed. We are a lot further along than we were, say 10 years ago, but there are still serious challenges with parts of the stack still not having good API access, humans are still doing too much manual work, etc.

Yes :) W/o automation "it" does not happen. But we have seen automation become a run away nightmare. Step one one troubleshooting for example - "turn off the automation"

Automation is in a good place for sure. #DevOps (esp. at scale) is barely possible without recent advances. But more still needed.

There will ALWAYS be more work needed in automation...but I do believe we are at an inflection point. Adoption will increase quickly as will power.

@jfryman The API point is a great one. Good blog that @cote pointed out in newsletter recently re: VMware's "API"s.

Automation is just the beginning. Operational process optimization takes more than just automation and it should be a stretch goal for devops if it isn't already.

e.g. infra, test, release automation are core to #DevOps; but automation is still lacking in N/W, dB, sec and esp. on enterprise platforms.

Automation is just the beginning. Operational process optimization takes more than just automation and it should be a stretch goal for devops if it isn't already.

@lmacvittie Your point on process optimization is important... Automating a task is quite different from automating a significant portion of a life cycle.

@jfryman do you think orchestration is fantasy til automation moves along or can it be parallel dev paths?

@lmacvittie Process optimization in Operations will be huge. Codifying process to abstract out patterns will be key. Eliminate unnecessary cruft while promoting actions with highest value return.

@3rdwaveinsights Yes yes YES. Processes are not tasks... but that distinction goes along with viewing ops as an integrated entity rather than as a discrete set of functions being managed ;-)

@3rdwaveinsights IOW, we're a ways off from getting to a systems view of ops. But at least we're on the right path...

@lmacvittie @3rdwaveinsights Got it, I think. @phool_stormer sometimes talks about open sourcing (via automation code) your ops patterns. Which suggests you better first know what they are. Similar thought?

@lmacvittie +1. I think that these actually play hand-in-hand. Good automation really looks at solving the problem based on the workflow, finding choke points, and eliminating the problem with automation. Process optimization is a natural next step.

can someone define "orchestration" in this #DevOps context? I'm curious what people think it is beyond/above "automation."

I think they happen in parallel. Automation really only focuses on the micro (task, at a domain level), and orchestration on the macro (tasks assembled to accomplish a unit of work).

having the workflow documented is the first step, and in my mind that constitutes a basic orchestration of sorts (manual + automatic steps) that then informs your priority/goals of automation

@cote I have a vague hierarchy in mind, task automation --> process automation --> orchestration. In #DevOps context, I see automation as within boundaries, orchestration between them. Not definitive, but how I think of it.

@lmacvittie As you said, we are way off getting a systems view...deeper discussions about the science behind "devops" rather than just culture is creating the right paht.

@AndiMann +1 I like that hierarchy, I think it logically extrapolates from one step to the next.

@AndiMann If we look to BPO and its evolution it's easier to see the progression. They've already done this... so has dev through Six Sigma...

depends on what you define as an Enterprise. if their "product" is digital, this is a growing segment for DevOps adoption.

Both! A bunch I know are doing straight up DevOps. But many are targeting SDLC flow constraints in just 1 or 2 areas, not calling it 'DevOps'

depends on what you define as an Enterprise. if their "product" is digital, this is a growing segment for DevOps adoption.

depends on what you define as an Enterprise. if their "product" is digital, this is a growing segment for DevOps adoption.

Whether that is DevOps? Does it matter? Ultimately dogma is not important, results are. Reminds me of Tom Bittman on cloud yrs ago - sometimes 80% is enough.

@AndiMann Nice. Key results = speed to respond to market (internal or external demand)? Or another better more crisp KPI?

many "enterprises" @451Research talks with are in the midst of figuring out how to apply "cloud" to making their IT, and business run better, and end-up looking to #DevOps as an appdev layer on-top of that.

I don't think we can stereotype an org as 'enterprise' anymore. Look at market leaders and see what they are calling it.

@epowell101 Rather than speed, agility. Doing something that used to be market relevant faster is quite different than jumping into something new fast

@3rdwaveinsights well, I think in this case speed is the enabler of agility. If the appdev cycle was 6 months, less agility would likely be achieved. The ability to quickly iterate and respond to feedback (good & bad) is the benefit of speed here.

@AndiMann agree that dogma clouds the issue (pun intended). It's about engineering the solutions having adaptive infra

@epowell101 Time to market for business ideas seems to be top driver & outcome. Cost, quality, cust. sat., revenue, penetration all critical, but speed tops the list I think.

@cote Cloud is a great enabler, but less of an answer to business problems, more a starting point. DevOps, Mobile, Social, Analytics, etc. are real solutions. Cloud is the platform for them.

@jakerobinson I disagree. Enterprise is not (should notbe) just a label; it indicates fundamentally different challenges - in scale, regulation, market, and more. imo

@cote Agree speed is an enabler... Getting feedback could also be viewed as an enabler... these all contribute to make the business more agile.

@AndiMann but saying 'what do enterprises call it' eludes that all enterprises move at the same speed and have the same culture. Just because an org is of a certain size does not mean they get there a certain way.

@AndiMann but saying 'what do enterprises call it' eludes that all enterprises move at the same speed and have the same culture. Just because an org is of a certain size does not mean they get there a certain way.

@AndiMann but saying 'what do enterprises call it' eludes that all enterprises move at the same speed and have the same culture. Just because an org is of a certain size does not mean they get there a certain way.

@AndiMann Excellent point, Andi. Enterprises have different challenges and thus different reasons for adopting DevOps, but they're just as valid - whatever they might call it.

coders who can engineer not just administrate. Node and integrated stacks; zero provisioning issues (manual) speed is key to success

...or the other way. Ops Eng who can code. It naturally lends itself to turning out DevOps oriented tool sets.

whether dev, QA, ops, manager, product mgr, etc.: people who are curious and want to explore new things. Few will ever have the "can do everything" utopia, but people can at least learn & block less to protect their comfy cube of ignorance.

@phool_stormer This is what i'm seeing as well; it's not about just Dev or Ops but engineering talent

SW Developers with an interest in Operations - everything from Application Ops to Deployment to Infrastructure

in the enterprise-y space, equally vital to "engineers" are making sure management and "business owners" are aligned. If you build a great appdev pipeline, and the money people don't care, it's likely a waste of your effort.

we can also look at how Agile Development played out in the 2000s for similar patterns and tips/tricks.

@cote True. I think cost savings, repeatability, and delivering business more quickly are great drivers for business owners.

@cote True. I think cost savings, repeatability, and delivering business more quickly are great drivers for business owners.

@cote True. I think cost savings, repeatability, and delivering business more quickly are great drivers for business owners.

@cote I think the latter point is key. The end goal is business enablement, not automation mecca. How do you find companies tackle the balance between the pipeline development and pipeline usage wrt automation?

@cote Yes! It can be quite an uphill battle getting adoption for DevOps tactics if management is not on board.

I'd argue that the Agile world crested at penetrating the business side of things, and that #DevOps has to be sure to "penetrate"/infect/turn the business to be long-term successful

There's a big difference between coding (an application) and coding (a script for automation or simple integration). That distinction needs to be made. We aren't talking full on app dev, we're talking scripting and targeted code.

There's a big difference between coding (an application) and coding (a script for automation or simple integration). That distinction needs to be made. We aren't talking full on app dev, we're talking scripting and targeted code.

@cote Agree with your penetration point.

@lmacvittie +1. Can scripting be a slippery slope though? We've seen orgs go through to refactor scripts in a way that basically turns them all together into a control plane. Is that just us? (self selection bias). Anyone else see it?

@epowell101 Absolutely can be a slippery slope. Discipline is necessary as is starting with design. It's not just the coding pieces of agile we need to adopt in ops.. it's the system-level approach up front we also need

@epowell101 Absolutely can be a slippery slope. Discipline is necessary as is starting with design. It's not just the coding pieces of agile we need to adopt in ops.. it's the system-level approach up front we also need

@epowell101 Absolutely can be a slippery slope. Discipline is necessary as is starting with design. It's not just the coding pieces of agile we need to adopt in ops.. it's the system-level approach up front we also need

@epowell101 Automating/scripting can be a slippery slope when the maintenance cost associated with lots of changes cause the ROIs to go negative.

@3rdwaveinsights Lots of changes to the automation itself? Do you get into the DevOps people, process, tools needed to author the automation for the DevOps people, process, tools?