The document discusses how IT teams often end up with multiple monitoring and management tools due to a lack of communication and trust between teams. This leads to an inefficient "architectural spaghetti" approach. It suggests adopting a DevOps approach to break down silos between teams and gain visibility into all tools and systems through a centralized monitoring platform to improve collaboration and efficiency. This can help avoid issues around lack of oversight, budgeting, and support that arise from a fragmented tooling approach.
24. SolarWinds, SolarWinds & Design, Orion, and Thwack are the exclusive property of
SolarWinds Worldwide, LLC or its affiliates, are registered with the U.S. Patent and Trademark
Office, and may be registered or pending registration in other countries. All other SolarWinds
trademarks, service marks, and logos may be common law marks or are registered or pending
registration. All other trademarks mentioned herein are used for identification purposes only
and are trademarks of (and may be registered trademarks) of their respective companies.
Notes de l'éditeur
Image: https://commons.wikimedia.org/wiki/File:Professors_cube.jpg
If your company had 27 different customer management systems, each one tailored to fit the needs of a distinct group of 5 individuals, would someone consider that to be an issue? (If you didn't know, at least one person should not like this idea, and that's the person writing the checks to purchase all these systems). OK, then why do we allow 27 other kinds of tools to be purchased? Like Monitoring.
But my point in telling you all this is that my examples will have a monitoring angle just because that's what I know and what I've seen. But it translates to almost any segment of IT life.
Icons are part of free library: http://www.opensecurityarchitecture.org/cms/library/icon-library
If you are anything like the 15 or so companies where I've worked, you might have an area of IT that looks like this. Maybe it's monitoring. Maybe it's versioning tools. Maybe it's documentation.
Now I'm not saying that "complexity bad" or "if you can't fit your design on a 3x5 notecard then it's probably over-engineered". The stuff that we do is non-trivial and sometimes it can get complex and still be good.
BUT... many times, the complexity didn't arrive by design, it happened organically.
Emojis free for use from here: http://emojione.com/
In IT complexity grows organically when people opt not to think too much about choices and changes being made.
There are two main reasons why: fear of conflict and lack of trust.
It's easier for a group to buy their own tool than to try to work with another group and share an existing system; or to go through the inevitable RFP process. It avoids being asked questions like
why didn't you include this in the budget request cycle
so I could cut it and watch your dreams crushed and become a hollow shell of a human being
why don't you use the corporate solution
which is non-existing, years out of date, or unsupported
why don't you put it up for RFP
are you frakking kidding me?
why don't you use the solution Bob's team found
because dual team ownership is a delightfully simple situation
also? Bob's team is the custodial staff. Not exactly an apples for apples situation
http://www.istockphoto.com/photo/angry-siblings-sitting-arms-crossed-with-sad-mother-on-sofa-gm479233710-64979783?st=_p_fight%20kids
(paid for)
Nobody is going to understand your shit like you do because, after all, it's your shit, and companies don't hire redundant teams doing the same exact job. So who ELSE is going to know how your shit works?
And that's key - how can there be a "one size fits all" or even "one size fits most" tool that will be everything you need. And if it's not everything you need, why would you sacrifice part of your hard-won budget for it?
And this is largely both a true statement and a good choice. FOR THE TEAM, IN THE SHORT TERM.
But it's NOT a good choice for the business, and often it's not a good choice for the team.
What I mean is that when the team picks a tool, other non-best-use-case factors become limitations
"how much can we spend without anyone noticing and/or needing to ask for approval"
"what will we know how to support"
"what will run on that spare laptop bob has under his desk" (or vm we didn't decomission, or within our AWS/Azure allowance)
Image: http://www.freeimages.com/photo/fire-extinguisher-1309526
Even when the tool is sanctioned by the company, if it is (as Alton Brown calls them) uni-tasker solution (ie: it is owned and used by your team but not the other teams) you run the risk of:
duplication of function
sometimes duplication of the SAME SOFTWARE in two departments
Image: https://pixabay.com/en/unicorn-mythical-creatures-mystical-611886/
Not by coincidence, but two of the dysfunctions that DevOps seeks to remove amongst teams are exactly that: trust issues and fear of conflict. Along with "too many alerts and too much noise" and "I don't have time to save time", these are the most common struggles we see within the DevOps structure amongst sys admins.
The first step in any solution is to admit you have a problem.
Image: https://pixabay.com/en/bowl-fish-goldfish-tank-water-sad-161409/
Can you even SEE the problem?
Do you have more servers than people?
Is the largest consumer of IT resources IT itself?
Does every team have it's own preferred tool?
Images:
https://commons.wikimedia.org/wiki/File:National_Data_Center_Capacity_Building_training_course_-_Flickr_-_The_Official_CTBTO_Photostream.jpg
https://pixabay.com/en/datacenter-servers-computers-286386/
It seems that everyone forgets how IT includes the word 'information'. It's the information you and your teams need, not the tools themselves.
How do we solve the problem of "there's an app for that" and tool sprawl? First and foremost, we stop focusing on the tools and instead focus on the data. Let the data and the analytics of the data be what drives your requirements, not a pretty GUI interface or customizations.
break down the elements
functions
primary, secondary, tertiary
where is there overlap
where is there a gap
when can you do "brick in/brick out"
show when intrinsic value is greater than the cost of the tool
Image: https://commons.wikimedia.org/wiki/File:Professors_cube.jpg
again, what I know is monitoring, so that's the examples I'm using YMMV, caveate emptor, blah blah.
monitoring is not management
ongoing monitoring is not emergency monitoring (smokeping)
Image: http://www.istockphoto.com/photo/young-business-children-make-faces-holding-lots-of-money-gm470201459-35093954
Paid for.
management speak
pull from buy me a pony
beware of words that don't mean the same thing
example: Log monitoring
example: APM (next slide)
If a person or group has a complaint and says they need to buy something new, turn that complaint into a request and see if you already have the data they need.
Use the data that exists and is already being collected to drive collaboration which will then build trust among your teams and remove the fear of conflict.