June 5, 2023 • 33 mins
Join Eveline Oehrlich and Luca Galante to discuss Platform Engineering.
Luca routinely speaks to tens of engineering teams every month. He summarizes his learnings and takeaways from looking at hundreds of DevOps setups into crisp, insightful reads for everyone in the industry, from beginner-Ops to cloud experts.
Luca’s additional resources:
Luca routinely speaks to tens of engineering teams every month. He summarizes his learnings and takeaways from looking at hundreds of DevOps setups into crisp, insightful reads for everyone in the industry, from beginner-Ops to cloud experts.
Luca’s additional resources:
- What is platform engineering?
- What is an internal developer platform?
- What is Dynamic Configuration Management?
- Platform Engineering community
- PlatformCon 2023
- Luca’s LinkedIn and Twitter
The Humans of DevOps Podcast is incredibly grateful to be voted one of the Best 25 DevOps Podcasts by Feedspot.
Want access to more DevOps-focused content and learning? Join SKILup IT Learning to gain access to a toolbox of hyper-relevant skills in a convenient, online learning platform focused entirely on DevOps and IT
Want access to more DevOps-focused content and learning? Join SKILup IT Learning to gain access to a toolbox of hyper-relevant skills in a convenient, online learning platform focused entirely on DevOps and IT
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Narrator (00:02):
You're listening to the humans of DevOps podcast, a
podcast focused on advancing thehumans of DevOps through skills,
knowledge, ideas, and learning,or the skil framework.
Luca Galante (00:16):
So usually, it's already quite a hurdle to
convince the executive and alldifferent internal stakeholders
to start and drive a platforminitiative that can and should
last for years.
Eveline Oehrlich (00:33):
Welcome to the humans of DevOps Podcast. I'm
Evelyn early Chief ResearchOfficer at DevOps Institute.
Today we have with us drumroll,Luca Galanti is a product owner.
And our title of our podcast isa conversation with a thought
leader on platform engineering,I just talked to look at finding
(00:56):
out from where he is coming toto us, I will not give it away.
But he's in a very beautifulplace. Hopefully, all of you
listening into us will be in abeautiful place, having a nice
cup of whatever in front of youenjoying our conversation. So
let me give you a little bit ofbackground on Luca, he actually
(01:18):
routinely speaks to 10s ofengineering teams every month,
tons of engineering teams, not10s. But tons. Every month, he
summarizes his learnings andtakeaways from looking at
hundreds of DevOps setups intovery crisp and insightful reads
for everyone in the industry.And that from beginner ops to
cloud experts. So Luke, welcometo our podcast.
Luca Galante (01:43):
Thank you. Thank you for having me. And hello to
everybody from Tenerife. Areally nice place.
Eveline Oehrlich (01:49):
Yeah, he gave it away, thank you give it away.
That is your choice of givingthat away. I've been to tenerife
myself and it is a beautifulplace. So enjoy it. Thank you
give us a little bit morebackground on your current
position and what you do Luca, totally
Luca Galante (02:09):
happy to so I run product at humanI tech. But I
guess I'm more known in thispace as one of the core
contributors to the platformengineering community. I
moderate the slack there wherewe have almost 10,000 People
actually I think we're goingacross 10,000 Next week, I host
platform con which is the firstever platform engineering
(02:31):
conference, I write a newsletterthat goes out to six or 7000
people. It's called Platformweekly, it goes out every week.
And yeah, so I talk a lot aboutplatform engineering. And as you
mentioned in the intro, alsokind of discuss what the
benefits can be with a lot ofsort of like infrastructure,
(02:52):
DevOps, and so on teams
Eveline Oehrlich (02:54):
all over the world. Fantastic. And I think
that's why of course, we arevery interested in talking to
you. Just again, for yourbenefit Luca, our audience is
really wide spectrum frombeginners, DevOps from more
advanced, we have practitioners,we've got site reliability
(03:14):
engineers, we probably got a fewplatform engineers on the call
on our on our podcasts listeningin, we've got testers, we've got
system administrators, we've gotleaders, we've got individual
contributors. So just so thatyou have a bit of a bit of a
background. Now, the termplatform engineering has gained
(03:36):
steam, and to the tune of folks,manual players. And Matthew
Skelton who I both know,mentioned, actually this topic
in their book, in 2019. I didsome research with a vendor
partner of the DevOps Institute,also in 2020, on site
reliability engineering, and wesaw the term show up there, then
(04:01):
as well. So you've been, as yousaid, a core contributor to this
platform engineering community.And we'll have later on for all
the listeners a more details onwhere you can get involved. But
before we go any further, Ialways love to make sure that we
have a definition. So what isthe definition of platform
(04:24):
engineering? Totally. So
Luca Galante (04:29):
the way I think about it is platform engineering
is really the art because it'smore of an art than a science,
in my opinion, of effectivelytaking all the tech and tools
that you have, especially in anenterprise engineer
organizations and binding theminto a golden path that enables
(04:50):
developers cell service and ingeneral reduces the cognitive
load for the individualcontributor and the sum of This
golden path that are built bythe platform engineering team
for the rest of the juniororganizations is often referred
to as an intern developer, orIDP for short. And the focus of
(05:14):
an IDP is really that ofimproving developer experience,
by lowering cognitive load todevelopers, while also
streamlining the kind of likeinfrastructure and operational
setup, on the other side ofthings. And a sort of like key
principle that we advocate forin the community is platform as
(05:35):
a product. And so really haveplatform teams focus on building
and shipping a platform as anyother product to the rest of the
G organizations, which is reallytheir internal customers and the
developers. And so that's kindof how we define platform
engineering, in turn developerplatform, which is the product
(05:59):
of platform engineering, and thetwo main actors in that
transaction, which is theplatform team, on the one hand,
apart from engineers, and thedevelopers, their internal
customers on the outer.
Eveline Oehrlich (06:12):
So IDP, that's something for everybody to do
note down. And so typically,where does such a team sit? If I
think of the organization,today, reporting to application
development, infrastructureoperations DevOps, and you
cannot say it depends on what'sthe typical. What's the typical
(06:34):
team these folks sit in, interms of the organization?
Luca Galante (06:41):
It's gonna be hard without the independence, about
answer. So, you know, thereality is, we see very
different kind of, sort of, likeorganizational designs,
depending on really the size.More importantly, right. So
you'll have teams of 100 200,engineers, where they report
(07:03):
directly to the CTO, you have,you know, come massive companies
that we work with 1000s ofengineers, and obviously, the
platform team does not reportdirectly to the seat to but to
some VP level, probablyresponsible on developer
experience, or cloudinfrastructure, or cloud ops.
You know, you have your kind of,like, Director of dev x had a
(07:25):
plot firm, those are the type oflike titles that usually lead
these initiatives.
Eveline Oehrlich (07:31):
So to some extent, I have also seen them as
a shared services group, ofcourse, that name is not really
Vogue, right? Who wants to be ashared services group, but I've
seen them as a as a as an addedvalue to, to that team, it kind
of depends on as you said, so Isaid, it depends. I didn't tell
(07:52):
myself, I was not supposed tosay that. So but thank you, that
was really, really good. Now,again, the goals of this team,
you alluded to it a little bitalready, but give us what are
their goals?
Luca Galante (08:06):
Totally. So the the main goal of the platform
team really is that of figuringout what the what is the minimum
common denominator, across theirentire engineer organizations
have problems that thedevelopers have, right? So you
(08:27):
really want to start from notOh, hey, let me through, like
build a platform with the latestshiny new tech, because I've
heard somewhere that it's cool.And I should, I should take a
look at that. But really take alook at you know, the complexity
and the reality of yourbrownfield enterprise setup,
where you usually have, youknow, many different development
(08:48):
teams with very different stacksand delivery setups, you know,
you have the ones that preferusing your circle CI applies
aren't go, the ones that are onJenkins, the other ones are
using something again. And soyou kind of want to take a look
at all of that and get anunderstanding, alright, what is
(09:08):
the minimum common denominatorthat souls the attacks, the
largest possible problemsurface, and that's really the
starting point of a platformteam, together with pinning down
precisely a mission and a rolefor the platform team that gets
marketed internally to the restof the organization, because
(09:31):
that's really important. Inorder to make sure that your
platform initiative issuccessful, is to make clear
that everybody makes sure thateverybody understands, hey, you
know, this is not anotherinfrastructure SRE team. This is
a product team. And so thisreconnects to what I was
mentioning in the beginning,which is really the core tenet
(09:53):
of platform engineering and whatclearly differentiates a
platform team from a DevOpsinfrastruc Archer essary team is
this product mindset, right? Sois the idea that you need to
build a very tight feedback loopwith the rest of the engineer
organizations, which again, areyour your customers, and make
(10:15):
sure that you iterate on top ofthis minimum common denominator
are problems that you identifiedat the beginning. And keep
shipping teachers and keepimproving the platform by
building different is differenttypes of golden paths that we
mentioned different in thedefinition. And this is really,
the I think the make or breakfor a successful platform team
(10:38):
is the ability of listening andcommunicating clearly,
internally, internally to theorganization. And understanding
what are the best golden pathsthat I can design for different
types of users across differenttypes of teams. Right. So as an
example, you'll have, you know,your senior backend engineer
that really enjoys handlingtheir Helm charts and icy
(11:03):
modules and some weird Yamofiles and scripts that they've
put together. If you, you know,build a platform that is very
you, hi, Avi, and completelyabstracts all of that away from
them, they'll fill, you know,they have no context, and
they'll push back on your pathfrom initiatives and that
(11:24):
platform initiative willprobably fail. On the flip side
of that you might have, youknow, a junior front end
engineer, that doesn't carewhether you're running on
Kubernetes, or somewhere else,they just want to deploy their
code, see, you know, test somechanges, and if you go and so
(11:45):
for them, it might make sense tohave a much higher level of
structuring, whether it'sthrough UI, or usually what we
find, on average is still youwant to do something code base,
because developers don't like tobe abstracted away into UI. But
that clearly shows you meet twodifferent types of golden paths,
and two different levels ofcontexts that you provide to
(12:06):
people. And, and so the keyability of a platform
engineering team, is that of isthat of? Sorry, is that a
listening and, you know, beingable to put together this
(12:27):
different golden paths for fordifferent apps to users?
Eveline Oehrlich (12:31):
You know, what it makes me think of a Rubik's
Cube when you were talking aboutthese examples, right? We have
very different types of paths asyou sit for these different
types of teams. So that's quitea challenge for for such a team
or that gets me to anotherquestion, or I'm assuming that
(12:54):
multiple platform teams arethere?
Luca Galante (12:59):
That's a great question. It depends, if I can
use that, sure, you know, sousually, in the vast majority of
cases, no, actually, you know,it's, it's already quite a
hurdle to get over to convince,you know, to get the executive
(13:19):
and all different internalstakeholders buy in, to start
and drive a platform initiativethat can and should last for
years. So usually, in the vastmajority organization, we
actually see one team, butobviously, the ideal scenario is
actually having multiple teamscompeting with each other. And
so we, I can, I can link in theshow notes there. I've spoken a
(13:43):
couple of times with ArneErickson is a also really key
contributor of the platformengineering community and other
platform engineer.org blog, andhe has a rolled out the what was
the successful platforminitiative of Salesforce about
five to eight years ago. And inour conversation, he tells a
(14:05):
story of how they workinternally with I believe three
or four outer different platformteams. And this you know,
highlights again, what I wastalking about earlier, which is
the importance of you know,communication and internal
marketing skills, which reallypositions the the platform
engineer to have a differentskill set and your kind of like
(14:27):
usual infrastructure SRE that ismore focus on you know,
reliability, scalability,maintenance, and so on. Because
really, you again, you areshipping a product and a key
part of shipping a product foreverybody that has done so well
know, is distribution, right,and building that relationship
to your customers. And sonowhere is that Sure. And then
(14:50):
in larger organizations whereyou're actually competing and so
you're basically building aalmost a go to market motion.
within your law firm design,
Narrator (15:04):
you want to get ahead in your career and we can help.
DevOps Institute certificationswill validate your knowledge and
give you the skills you need tosucceed in a key area of DevOps.
With our certifications, you'llbe able to prove your subject
matter expertise, and enhanceyour professional credibility.
advance your career today withDevOps Institute certifications,
(15:25):
get started by visiting DevOpsinstitute.com.
Eveline Oehrlich (15:31):
So you mentioned already a variety of
benefits of platform engineeringteam, let's not dwell further on
that one. Because I think thereis plenty of good things you
already said. But challenges andyour when you mentioned one
already, which is the actualmaking sure that executive
spying? So two prong questionfirst, what are some other
(15:55):
challenges besides making surethat you market that? And then
how and what is the best way toactually market that if I'm the
CIO of company X, and you arenow coming to me? How would you
market that?
Luca Galante (16:10):
To me? Yeah, that's a great question. So I
think the you know, there aretwo basically set of challenges.
One is a, you know, set oftechnical challenges. And again,
I think the, the best sort oflike starting point is looking
at the current reality of yourengineer organizations and find
(16:35):
that lowest common techdenominator that can be used
across all different teams andstart from there. And that by
the way, just as a quicktangent, is something that we
see a lot of people miss, as aas an idea. Because I feel like
the fallacy is a lot of peoplethink, hey, if I could only
(16:55):
start Greenfield, right, and soI'm a new platform teams and
your platform initiative, let'sstart from scratch, right. But
by starting from scratch, andkind of dreaming up of your
ideal bathroom stack, you aremissing out on a key piece of
the puzzle, which is listeningand, and effectively building on
top of the information that youalready have available. And so
(17:19):
actually, almost paradoxically,we've see the most successful
part from initiatives really arein very complex enterprise
setups, there are very, that arevery sort of nuanced, and the in
the different stacks that peopleuse, and then learn and build on
(17:40):
top of that, and then and thenfrom their, from their they get
started. The other set ofchallenges is what we mentioned,
which is basically internalmarketing. And then there there
are, there are two, I thinkdifferent levels. One is the
first one that you mentioned,which is basically you know,
your your kind of like executivebuy in. And I think the main
(18:05):
line of argumentation there isone around ROI. And especially
in you know, times like now andthe macro situations that we're
living through is paramount, Ithink for engineering management
to understand the impact that aplatform can have. And the
(18:25):
impact is not only on your kindof Dora metrics, right? So
obviously your lead timedeployment frequency, what is it
change failure rate and MTTR?They all improve? Right?
Massively, you have a crazy dropin OPS overhead in ticket what
we call ticket ops, right? Sobasically, operations team
(18:47):
constantly being stuck. And thenon the other side of the fence,
developers having to wait so youknow, waiting times drop ups,
overhead drops, you also haveyour kind of slo metrics,
improving like security,stability, reliability, all
improved because you get a muchbetter audit of who deploys
(19:08):
what, where you can easily dodis rollbacks between
deployments. And so your entiredelivery setup gets extremely
more efficient. But I think theother thing that sometimes can
be a little bit, maybe harder,like it's a little bit less
quantitative, potentially, butit's the, you know, crazy drop
(19:28):
in frustration across theengineer organization, right?
Because in a lot of in a lot ofteams right now, the reality of
DevOps is developers beingcompletely overwhelmed by
increasingly complex cloudnative tool chains that adults
fully understand and in factshouldn't because it's not their
(19:49):
job. And so when they just wantto, you know, spin up a pre
environment to test a littlechanged is now all of a sudden
need to touch 1015 differenttools and three or four
different scripts andcompletely, are completely
overwhelmed by what we call, youknow, this cognitive overload.
And, and on the flip side ofthat, so what happens is then
(20:11):
they basically either ping theirmost senior colleague, that
senior backend engineer wementioned earlier that likes,
you know, messing around withtheir Yamo files, maybe. But
then what you end up having isyour most senior and most
expensive resources, shall Isay, being effectively blocked
and blocking outers and becominga bottleneck. So that's another
(20:31):
point for your kind of like sealevel conversation. Yeah. And
then on the flip, thealternative with that, again, as
you're like infrastructure andops team completely stuck,
basically constantly fightingticket ops and putting out
fires. And so there's reallywidespread frustration on both
sides of the aisle. And that'sreally what platform engineering
sets out to improve is reallyliving and enabling true DevOps,
(20:59):
true, you build it, you run it,and not the kind of like half
baked version that we ended upwith in the last decade,
effectively. And so that'sreally kind of like your pitch
to sea level. And obviously, youknow, you can, you can take
smaller parts of that and gopitch to your developers and
(21:20):
operations team on the other.After that, I think what's
important there on theoperations side of things is
that you make really clearagain, that you you're
differentiating between sort ofmore traditional operation role,
right. So it's not like aplatform team replaces an SRE
team, you still need to careabout how your load out your
(21:45):
load balancing across all yourproduction regions, and
clusters, and all that goodstuff. But you need to spin up
this separate set of skill setsand roles that really focus on
shipping a product and you needto make sure that that is really
aligned the mission. And thevision of the platform
engineering org is aligned withthe rest of the serve like
(22:09):
infrastructure teams to makesure that there are no friction,
there's no friction there. Andthen the third sort of group of
stakeholders, obviously, asdevelopers, and we kind of
touched on this earlier, but theidea here is, you just need to
listen a lot and make sure thatdevelopers have a clear idea
(22:31):
that you are improving their dayto day math and lead by enabling
them to self serve the tech andtools they need to run and
operate their operations infulsome service completely
independently. But at the sametime, you're not doing that by
providing them some past like,some class like tool that
(22:53):
abstracts them away from theunderlying tech and tools. And
again, this will matter more orless, depending on the user. If
it's a more senior user that is,you know, used to handling low
code stuff then. Or just likelow level configuration stuff,
they will be very sensitive tothe type of golden path that you
(23:17):
design for them. And that'swhere it's really key that that
you know that that informationflow is really optimized. And
then also, finally, I thinkthere is a conversational roll
up strategy. Again, you canthink of it as any sort of like
go to market, who are your earlyadopters, where you start. And I
(23:37):
think, you know, we've seen alot of successful platform
initiatives starting by workingwith the more advanced
development teams that usuallyare the ones who are going to be
more comfortable with your cloudnative stack, you know, they're
not afraid of Helm charts, orTerraForm, modules, and so on.
And so it will be easier tostart by sort of like with them,
(24:02):
laying that foundation withsomething that is not so
abstracted, and thenprogressively move to maybe more
junior or less experienced teamsor just teams that frankly,
don't need to be so deeply intouch with the underlying stack
of things. And for them build apotentially more abstracted,
(24:24):
simpler golden path. So thosethree groups,
Eveline Oehrlich (24:28):
I can just hear your passion on this topic
is just phenomenal. Luca, youare phenomenal. I want to say
that one more time. You arephenomenal. Now I have two more
questions. One is related to theplatform engineering and that
individual skills. And the lastone is a fun question. So let's
do the hard work first. skillsas you know, we hear the DevOps
(24:51):
Institute are really engaged inskill building upskilling and
all those wonderful topics andyou mentioned one Skill already,
which was listening something Ihave to tell my husband, he
should become maybe a platformengineer, and then he would
improve his listening skills.But no, he has no chance. But
that was one skill. You said,what are some additional two,
(25:15):
maybe two other skills you wantto highlight? If I want to turn
into a platform engineer, whatshould I have as a skill?
Luca Galante (25:25):
Totally well, so I mean, from a technical
perspective, obviously, you wantto have a solid understanding of
all your kind of, you know,infrastructure, especially cloud
native tool chain, right. Soyour Kubernetes I see
infrastructures code, see ICDworkflows, good ops, you know,
throw your boss word of dude,you were, you know, obviously,
(25:49):
you wanna, you want to havethose bases covered. But I think
the vast majority of people thatwill approach to platform
engineering, sort of topicscoming from a, kind of like,
your infrastructure DevOps rolewill already be familiar with
that. I think, really, and Iwant to kind of de risk of
repeating myself, but I thinkthat there are two main things
(26:12):
really, that I see. One isproduct mindset. And so that's
really the shift that maybe ifyou are coming, sort of like,
more green to this is, or,again, from building and
shipping products, that's not abig shift. But I think if you
aren't coming from your kind ofmore traditional infrastructure
(26:34):
and DevOps role, it might be amore important shift, mindset
wise that you need to make. And,and that's, I think, really key,
right? And, and so that'sobviously like the part of the
listening and having thismindset and really starting to
think, Hey, your developers areyour customers. And that's all
(26:57):
that matters. And so thenobviously, there are oldest
topics that have emerged fromthere. The the kind of
fundamentals of productmanagement, right, that we all
know, so user research product,roadmap, MVP being, how you
think about adoption, and that'sjust to mention a few, right? So
really just taking everythingwe've learned in the last few
decades around productmanagement and applying it to
(27:19):
effectively infrastructure andgluing together your tool chain
into a product, a platform thatmakes sense for your end users.
And then the other one iscommunication broadly, and, you
know, obviously, part of visionis listening. But the other part
is speaking. And so the speakingpart is again, like how do you
think about internal marketing?How do you think about
(27:42):
stakeholder buy in? And how doyou make sure that you are
adapting to the different typeof lingo to communicate
incentives clearly, right. Sothe way you communicate to sea
level, and you talk about ROI,and you talk about, you know,
your decrease time to market,you know, developers don't care
(28:04):
about that. Right. So then whenyou talk to developers, it's
really about, you know, theirdeveloper, velocity, I mean,
even velocity productivity ormore stuff for sea level, it's
really about just getting themstuck and talking about, Hey,
your waiting times will drop.And, you know, this is how this
is how we can improve your dayto day immediately. And I think
(28:29):
they're a key. A key part of thecommunication is how you plug
into their current workflow, wethought without interrupting.
And so that's, I think, isreally important. But yeah,
those those two things, reallyproduct mindset and
communication skills are, Ithink, the major thing and if
(28:49):
you were to plot on a graph,your kind of evolution of that
buzzworthy sort of roles andtitles from your SIS admin, to
your infrastructure to yourdoubts in the near future, apart
from engineer, you know, if youhad time on the x axis and sort
of like communication on the yaxis, then you know, they would
(29:10):
plot on to the right, andthat's, that's, I think, kind of
how you want to think about it,right? Like platform engineer is
basically a DevOps engineer thatlistens and communicates better.
Eveline Oehrlich (29:21):
Beautiful. Wow, that sounds like a really
exciting career choice to makeif anybody out there wants to
know more, there is plenty moreto get involved in. I just say
that there is a fantastic reportand I think you will look about
the author of it. It's calledthe state of platform
engineering. So if anybody isinterested in then just look for
(29:43):
Luca Galanti and type in stateof platform engineering, because
there is all these differentways of getting involved in
Slack channels and communitiesand events. Super. This was
extremely insightful. I think.Now Now comes the fun part. So
besides doing these greatthought leadership pieces, doing
(30:05):
your work, talking to peoplelike me and platform
engineering, what do you do forfun Luca?
Luca Galante (30:13):
Um, well, so as I mentioned at the beginning, I'm
currently on Tenerife. And soit's been really fun over here.
I've been living out of eventactually, for the last couple of
weeks and surfing mostly, reallyearly in the morning, because
we're one hour behind here. Butthat's, I think, one of the
(30:33):
things that I do for fun, youknow, just just being in the
water being in nature, that's,that's really good times for me.
And the other one, frankly, as agood Italian is just eating and
drinking good wine. So, youknow, simple, simple person
simple needs, but does go prettyfar for me.
Eveline Oehrlich (30:51):
Fantastic. We have to meet in person, I have a
van. I love Italian food. Oh,amazing. I cannot serve you with
so I do like the water. So maybewe get to meet one time
somewhere in a place. MaybeJennifer, this has been
fantastic. Thank you. We've beentalking to Luke. I like Galanti
product owner, product master,and really a great thought
(31:15):
leader on platform engineering.Luca, thank you again so much
for joining me today on humansof DevOps podcast.
Luca Galante (31:23):
Thank you so much for having me. And if they can,
if I may, just because youmentioned the the events. If
people want to learn more aboutplatform engineering, I would
encourage them to check outplatform eg.org. And that's
really the home of the communityand platform khan.com. That's
the event that's coming up inJune this year. It's free,
(31:45):
anybody can sign up and we'llhave the best of the best really
of DevOps and platformengineering thought leaders
there. So I invite everybody tojoin us over there.
Eveline Oehrlich (31:56):
Great, thank you, humans. Yes, humans of
DevOps podcast is produced byDevOps Institute. Our audio
production team includes Juliapape, Daniel Newman, Schultz and
Brendon lay. Those are the threefolks I want to have a shout out
to because they're wonderful. Iam humans of DevOps podcast,
(32:16):
executive producer Evelynearlyish. If you would like to
join us on a podcast, pleasecontact us at humans have dev
ops podcast at DevOpsinstitute.com. Do I sound like a
virtual agent? Maybe? Anyway,have a great day. I'm Emily,
talk to you soon.
Narrator (32:39):
Thanks for listening to this episode of the humans of
DevOps podcast. Don't forget tojoin our global community to get
access to even more greatresources like this. Until next
time, remember, you are part ofsomething bigger than yourself.
You belong
You're listening to the humans of DevOps podcast, a
podcast focused on advancing thehumans of DevOps through skills,
knowledge, ideas, and learning,or the skil framework.
Luca Galante (00:16):
So usually, it's already quite a hurdle to
convince the executive and alldifferent internal stakeholders
to start and drive a platforminitiative that can and should
last for years.
Eveline Oehrlich (00:33):
Welcome to the humans of DevOps Podcast. I'm
Evelyn early Chief ResearchOfficer at DevOps Institute.
Today we have with us drumroll,Luca Galanti is a product owner.
And our title of our podcast isa conversation with a thought
leader on platform engineering,I just talked to look at finding
(00:56):
out from where he is coming toto us, I will not give it away.
But he's in a very beautifulplace. Hopefully, all of you
listening into us will be in abeautiful place, having a nice
cup of whatever in front of youenjoying our conversation. So
let me give you a little bit ofbackground on Luca, he actually
(01:18):
routinely speaks to 10s ofengineering teams every month,
tons of engineering teams, not10s. But tons. Every month, he
summarizes his learnings andtakeaways from looking at
hundreds of DevOps setups intovery crisp and insightful reads
for everyone in the industry.And that from beginner ops to
cloud experts. So Luke, welcometo our podcast.
Luca Galante (01:43):
Thank you. Thank you for having me. And hello to
everybody from Tenerife. Areally nice place.
Eveline Oehrlich (01:49):
Yeah, he gave it away, thank you give it away.
That is your choice of givingthat away. I've been to tenerife
myself and it is a beautifulplace. So enjoy it. Thank you
give us a little bit morebackground on your current
position and what you do Luca, totally
Luca Galante (02:09):
happy to so I run product at humanI tech. But I
guess I'm more known in thispace as one of the core
contributors to the platformengineering community. I
moderate the slack there wherewe have almost 10,000 People
actually I think we're goingacross 10,000 Next week, I host
platform con which is the firstever platform engineering
(02:31):
conference, I write a newsletterthat goes out to six or 7000
people. It's called Platformweekly, it goes out every week.
And yeah, so I talk a lot aboutplatform engineering. And as you
mentioned in the intro, alsokind of discuss what the
benefits can be with a lot ofsort of like infrastructure,
(02:52):
DevOps, and so on teams
Eveline Oehrlich (02:54):
all over the world. Fantastic. And I think
that's why of course, we arevery interested in talking to
you. Just again, for yourbenefit Luca, our audience is
really wide spectrum frombeginners, DevOps from more
advanced, we have practitioners,we've got site reliability
(03:14):
engineers, we probably got a fewplatform engineers on the call
on our on our podcasts listeningin, we've got testers, we've got
system administrators, we've gotleaders, we've got individual
contributors. So just so thatyou have a bit of a bit of a
background. Now, the termplatform engineering has gained
(03:36):
steam, and to the tune of folks,manual players. And Matthew
Skelton who I both know,mentioned, actually this topic
in their book, in 2019. I didsome research with a vendor
partner of the DevOps Institute,also in 2020, on site
reliability engineering, and wesaw the term show up there, then
(04:01):
as well. So you've been, as yousaid, a core contributor to this
platform engineering community.And we'll have later on for all
the listeners a more details onwhere you can get involved. But
before we go any further, Ialways love to make sure that we
have a definition. So what isthe definition of platform
(04:24):
engineering? Totally. So
Luca Galante (04:29):
the way I think about it is platform engineering
is really the art because it'smore of an art than a science,
in my opinion, of effectivelytaking all the tech and tools
that you have, especially in anenterprise engineer
organizations and binding theminto a golden path that enables
(04:50):
developers cell service and ingeneral reduces the cognitive
load for the individualcontributor and the sum of This
golden path that are built bythe platform engineering team
for the rest of the juniororganizations is often referred
to as an intern developer, orIDP for short. And the focus of
(05:14):
an IDP is really that ofimproving developer experience,
by lowering cognitive load todevelopers, while also
streamlining the kind of likeinfrastructure and operational
setup, on the other side ofthings. And a sort of like key
principle that we advocate forin the community is platform as
(05:35):
a product. And so really haveplatform teams focus on building
and shipping a platform as anyother product to the rest of the
G organizations, which is reallytheir internal customers and the
developers. And so that's kindof how we define platform
engineering, in turn developerplatform, which is the product
(05:59):
of platform engineering, and thetwo main actors in that
transaction, which is theplatform team, on the one hand,
apart from engineers, and thedevelopers, their internal
customers on the outer.
Eveline Oehrlich (06:12):
So IDP, that's something for everybody to do
note down. And so typically,where does such a team sit? If I
think of the organization,today, reporting to application
development, infrastructureoperations DevOps, and you
cannot say it depends on what'sthe typical. What's the typical
(06:34):
team these folks sit in, interms of the organization?
Luca Galante (06:41):
It's gonna be hard without the independence, about
answer. So, you know, thereality is, we see very
different kind of, sort of, likeorganizational designs,
depending on really the size.More importantly, right. So
you'll have teams of 100 200,engineers, where they report
(07:03):
directly to the CTO, you have,you know, come massive companies
that we work with 1000s ofengineers, and obviously, the
platform team does not reportdirectly to the seat to but to
some VP level, probablyresponsible on developer
experience, or cloudinfrastructure, or cloud ops.
You know, you have your kind of,like, Director of dev x had a
(07:25):
plot firm, those are the type oflike titles that usually lead
these initiatives.
Eveline Oehrlich (07:31):
So to some extent, I have also seen them as
a shared services group, ofcourse, that name is not really
Vogue, right? Who wants to be ashared services group, but I've
seen them as a as a as an addedvalue to, to that team, it kind
of depends on as you said, so Isaid, it depends. I didn't tell
(07:52):
myself, I was not supposed tosay that. So but thank you, that
was really, really good. Now,again, the goals of this team,
you alluded to it a little bitalready, but give us what are
their goals?
Luca Galante (08:06):
Totally. So the the main goal of the platform
team really is that of figuringout what the what is the minimum
common denominator, across theirentire engineer organizations
have problems that thedevelopers have, right? So you
(08:27):
really want to start from notOh, hey, let me through, like
build a platform with the latestshiny new tech, because I've
heard somewhere that it's cool.And I should, I should take a
look at that. But really take alook at you know, the complexity
and the reality of yourbrownfield enterprise setup,
where you usually have, youknow, many different development
(08:48):
teams with very different stacksand delivery setups, you know,
you have the ones that preferusing your circle CI applies
aren't go, the ones that are onJenkins, the other ones are
using something again. And soyou kind of want to take a look
at all of that and get anunderstanding, alright, what is
(09:08):
the minimum common denominatorthat souls the attacks, the
largest possible problemsurface, and that's really the
starting point of a platformteam, together with pinning down
precisely a mission and a rolefor the platform team that gets
marketed internally to the restof the organization, because
(09:31):
that's really important. Inorder to make sure that your
platform initiative issuccessful, is to make clear
that everybody makes sure thateverybody understands, hey, you
know, this is not anotherinfrastructure SRE team. This is
a product team. And so thisreconnects to what I was
mentioning in the beginning,which is really the core tenet
(09:53):
of platform engineering and whatclearly differentiates a
platform team from a DevOpsinfrastruc Archer essary team is
this product mindset, right? Sois the idea that you need to
build a very tight feedback loopwith the rest of the engineer
organizations, which again, areyour your customers, and make
(10:15):
sure that you iterate on top ofthis minimum common denominator
are problems that you identifiedat the beginning. And keep
shipping teachers and keepimproving the platform by
building different is differenttypes of golden paths that we
mentioned different in thedefinition. And this is really,
the I think the make or breakfor a successful platform team
(10:38):
is the ability of listening andcommunicating clearly,
internally, internally to theorganization. And understanding
what are the best golden pathsthat I can design for different
types of users across differenttypes of teams. Right. So as an
example, you'll have, you know,your senior backend engineer
that really enjoys handlingtheir Helm charts and icy
(11:03):
modules and some weird Yamofiles and scripts that they've
put together. If you, you know,build a platform that is very
you, hi, Avi, and completelyabstracts all of that away from
them, they'll fill, you know,they have no context, and
they'll push back on your pathfrom initiatives and that
(11:24):
platform initiative willprobably fail. On the flip side
of that you might have, youknow, a junior front end
engineer, that doesn't carewhether you're running on
Kubernetes, or somewhere else,they just want to deploy their
code, see, you know, test somechanges, and if you go and so
(11:45):
for them, it might make sense tohave a much higher level of
structuring, whether it'sthrough UI, or usually what we
find, on average is still youwant to do something code base,
because developers don't like tobe abstracted away into UI. But
that clearly shows you meet twodifferent types of golden paths,
and two different levels ofcontexts that you provide to
(12:06):
people. And, and so the keyability of a platform
engineering team, is that of isthat of? Sorry, is that a
listening and, you know, beingable to put together this
(12:27):
different golden paths for fordifferent apps to users?
Eveline Oehrlich (12:31):
You know, what it makes me think of a Rubik's
Cube when you were talking aboutthese examples, right? We have
very different types of paths asyou sit for these different
types of teams. So that's quitea challenge for for such a team
or that gets me to anotherquestion, or I'm assuming that
(12:54):
multiple platform teams arethere?
Luca Galante (12:59):
That's a great question. It depends, if I can
use that, sure, you know, sousually, in the vast majority of
cases, no, actually, you know,it's, it's already quite a
hurdle to get over to convince,you know, to get the executive
(13:19):
and all different internalstakeholders buy in, to start
and drive a platform initiativethat can and should last for
years. So usually, in the vastmajority organization, we
actually see one team, butobviously, the ideal scenario is
actually having multiple teamscompeting with each other. And
so we, I can, I can link in theshow notes there. I've spoken a
(13:43):
couple of times with ArneErickson is a also really key
contributor of the platformengineering community and other
platform engineer.org blog, andhe has a rolled out the what was
the successful platforminitiative of Salesforce about
five to eight years ago. And inour conversation, he tells a
(14:05):
story of how they workinternally with I believe three
or four outer different platformteams. And this you know,
highlights again, what I wastalking about earlier, which is
the importance of you know,communication and internal
marketing skills, which reallypositions the the platform
engineer to have a differentskill set and your kind of like
(14:27):
usual infrastructure SRE that ismore focus on you know,
reliability, scalability,maintenance, and so on. Because
really, you again, you areshipping a product and a key
part of shipping a product foreverybody that has done so well
know, is distribution, right,and building that relationship
to your customers. And sonowhere is that Sure. And then
(14:50):
in larger organizations whereyou're actually competing and so
you're basically building aalmost a go to market motion.
within your law firm design,
Narrator (15:04):
you want to get ahead in your career and we can help.
DevOps Institute certificationswill validate your knowledge and
give you the skills you need tosucceed in a key area of DevOps.
With our certifications, you'llbe able to prove your subject
matter expertise, and enhanceyour professional credibility.
advance your career today withDevOps Institute certifications,
(15:25):
get started by visiting DevOpsinstitute.com.
Eveline Oehrlich (15:31):
So you mentioned already a variety of
benefits of platform engineeringteam, let's not dwell further on
that one. Because I think thereis plenty of good things you
already said. But challenges andyour when you mentioned one
already, which is the actualmaking sure that executive
spying? So two prong questionfirst, what are some other
(15:55):
challenges besides making surethat you market that? And then
how and what is the best way toactually market that if I'm the
CIO of company X, and you arenow coming to me? How would you
market that?
Luca Galante (16:10):
To me? Yeah, that's a great question. So I
think the you know, there aretwo basically set of challenges.
One is a, you know, set oftechnical challenges. And again,
I think the, the best sort oflike starting point is looking
at the current reality of yourengineer organizations and find
(16:35):
that lowest common techdenominator that can be used
across all different teams andstart from there. And that by
the way, just as a quicktangent, is something that we
see a lot of people miss, as aas an idea. Because I feel like
the fallacy is a lot of peoplethink, hey, if I could only
(16:55):
start Greenfield, right, and soI'm a new platform teams and
your platform initiative, let'sstart from scratch, right. But
by starting from scratch, andkind of dreaming up of your
ideal bathroom stack, you aremissing out on a key piece of
the puzzle, which is listeningand, and effectively building on
top of the information that youalready have available. And so
(17:19):
actually, almost paradoxically,we've see the most successful
part from initiatives really arein very complex enterprise
setups, there are very, that arevery sort of nuanced, and the in
the different stacks that peopleuse, and then learn and build on
(17:40):
top of that, and then and thenfrom their, from their they get
started. The other set ofchallenges is what we mentioned,
which is basically internalmarketing. And then there there
are, there are two, I thinkdifferent levels. One is the
first one that you mentioned,which is basically you know,
your your kind of like executivebuy in. And I think the main
(18:05):
line of argumentation there isone around ROI. And especially
in you know, times like now andthe macro situations that we're
living through is paramount, Ithink for engineering management
to understand the impact that aplatform can have. And the
(18:25):
impact is not only on your kindof Dora metrics, right? So
obviously your lead timedeployment frequency, what is it
change failure rate and MTTR?They all improve? Right?
Massively, you have a crazy dropin OPS overhead in ticket what
we call ticket ops, right? Sobasically, operations team
(18:47):
constantly being stuck. And thenon the other side of the fence,
developers having to wait so youknow, waiting times drop ups,
overhead drops, you also haveyour kind of slo metrics,
improving like security,stability, reliability, all
improved because you get a muchbetter audit of who deploys
(19:08):
what, where you can easily dodis rollbacks between
deployments. And so your entiredelivery setup gets extremely
more efficient. But I think theother thing that sometimes can
be a little bit, maybe harder,like it's a little bit less
quantitative, potentially, butit's the, you know, crazy drop
(19:28):
in frustration across theengineer organization, right?
Because in a lot of in a lot ofteams right now, the reality of
DevOps is developers beingcompletely overwhelmed by
increasingly complex cloudnative tool chains that adults
fully understand and in factshouldn't because it's not their
(19:49):
job. And so when they just wantto, you know, spin up a pre
environment to test a littlechanged is now all of a sudden
need to touch 1015 differenttools and three or four
different scripts andcompletely, are completely
overwhelmed by what we call, youknow, this cognitive overload.
And, and on the flip side ofthat, so what happens is then
(20:11):
they basically either ping theirmost senior colleague, that
senior backend engineer wementioned earlier that likes,
you know, messing around withtheir Yamo files, maybe. But
then what you end up having isyour most senior and most
expensive resources, shall Isay, being effectively blocked
and blocking outers and becominga bottleneck. So that's another
(20:31):
point for your kind of like sealevel conversation. Yeah. And
then on the flip, thealternative with that, again, as
you're like infrastructure andops team completely stuck,
basically constantly fightingticket ops and putting out
fires. And so there's reallywidespread frustration on both
sides of the aisle. And that'sreally what platform engineering
sets out to improve is reallyliving and enabling true DevOps,
(20:59):
true, you build it, you run it,and not the kind of like half
baked version that we ended upwith in the last decade,
effectively. And so that'sreally kind of like your pitch
to sea level. And obviously, youknow, you can, you can take
smaller parts of that and gopitch to your developers and
(21:20):
operations team on the other.After that, I think what's
important there on theoperations side of things is
that you make really clearagain, that you you're
differentiating between sort ofmore traditional operation role,
right. So it's not like aplatform team replaces an SRE
team, you still need to careabout how your load out your
(21:45):
load balancing across all yourproduction regions, and
clusters, and all that goodstuff. But you need to spin up
this separate set of skill setsand roles that really focus on
shipping a product and you needto make sure that that is really
aligned the mission. And thevision of the platform
engineering org is aligned withthe rest of the serve like
(22:09):
infrastructure teams to makesure that there are no friction,
there's no friction there. Andthen the third sort of group of
stakeholders, obviously, asdevelopers, and we kind of
touched on this earlier, but theidea here is, you just need to
listen a lot and make sure thatdevelopers have a clear idea
(22:31):
that you are improving their dayto day math and lead by enabling
them to self serve the tech andtools they need to run and
operate their operations infulsome service completely
independently. But at the sametime, you're not doing that by
providing them some past like,some class like tool that
(22:53):
abstracts them away from theunderlying tech and tools. And
again, this will matter more orless, depending on the user. If
it's a more senior user that is,you know, used to handling low
code stuff then. Or just likelow level configuration stuff,
they will be very sensitive tothe type of golden path that you
(23:17):
design for them. And that'swhere it's really key that that
you know that that informationflow is really optimized. And
then also, finally, I thinkthere is a conversational roll
up strategy. Again, you canthink of it as any sort of like
go to market, who are your earlyadopters, where you start. And I
(23:37):
think, you know, we've seen alot of successful platform
initiatives starting by workingwith the more advanced
development teams that usuallyare the ones who are going to be
more comfortable with your cloudnative stack, you know, they're
not afraid of Helm charts, orTerraForm, modules, and so on.
And so it will be easier tostart by sort of like with them,
(24:02):
laying that foundation withsomething that is not so
abstracted, and thenprogressively move to maybe more
junior or less experienced teamsor just teams that frankly,
don't need to be so deeply intouch with the underlying stack
of things. And for them build apotentially more abstracted,
(24:24):
simpler golden path. So thosethree groups,
Eveline Oehrlich (24:28):
I can just hear your passion on this topic
is just phenomenal. Luca, youare phenomenal. I want to say
that one more time. You arephenomenal. Now I have two more
questions. One is related to theplatform engineering and that
individual skills. And the lastone is a fun question. So let's
do the hard work first. skillsas you know, we hear the DevOps
(24:51):
Institute are really engaged inskill building upskilling and
all those wonderful topics andyou mentioned one Skill already,
which was listening something Ihave to tell my husband, he
should become maybe a platformengineer, and then he would
improve his listening skills.But no, he has no chance. But
that was one skill. You said,what are some additional two,
(25:15):
maybe two other skills you wantto highlight? If I want to turn
into a platform engineer, whatshould I have as a skill?
Luca Galante (25:25):
Totally well, so I mean, from a technical
perspective, obviously, you wantto have a solid understanding of
all your kind of, you know,infrastructure, especially cloud
native tool chain, right. Soyour Kubernetes I see
infrastructures code, see ICDworkflows, good ops, you know,
throw your boss word of dude,you were, you know, obviously,
(25:49):
you wanna, you want to havethose bases covered. But I think
the vast majority of people thatwill approach to platform
engineering, sort of topicscoming from a, kind of like,
your infrastructure DevOps rolewill already be familiar with
that. I think, really, and Iwant to kind of de risk of
repeating myself, but I thinkthat there are two main things
(26:12):
really, that I see. One isproduct mindset. And so that's
really the shift that maybe ifyou are coming, sort of like,
more green to this is, or,again, from building and
shipping products, that's not abig shift. But I think if you
aren't coming from your kind ofmore traditional infrastructure
(26:34):
and DevOps role, it might be amore important shift, mindset
wise that you need to make. And,and that's, I think, really key,
right? And, and so that'sobviously like the part of the
listening and having thismindset and really starting to
think, Hey, your developers areyour customers. And that's all
(26:57):
that matters. And so thenobviously, there are oldest
topics that have emerged fromthere. The the kind of
fundamentals of productmanagement, right, that we all
know, so user research product,roadmap, MVP being, how you
think about adoption, and that'sjust to mention a few, right? So
really just taking everythingwe've learned in the last few
decades around productmanagement and applying it to
(27:19):
effectively infrastructure andgluing together your tool chain
into a product, a platform thatmakes sense for your end users.
And then the other one iscommunication broadly, and, you
know, obviously, part of visionis listening. But the other part
is speaking. And so the speakingpart is again, like how do you
think about internal marketing?How do you think about
(27:42):
stakeholder buy in? And how doyou make sure that you are
adapting to the different typeof lingo to communicate
incentives clearly, right. Sothe way you communicate to sea
level, and you talk about ROI,and you talk about, you know,
your decrease time to market,you know, developers don't care
(28:04):
about that. Right. So then whenyou talk to developers, it's
really about, you know, theirdeveloper, velocity, I mean,
even velocity productivity ormore stuff for sea level, it's
really about just getting themstuck and talking about, Hey,
your waiting times will drop.And, you know, this is how this
is how we can improve your dayto day immediately. And I think
(28:29):
they're a key. A key part of thecommunication is how you plug
into their current workflow, wethought without interrupting.
And so that's, I think, isreally important. But yeah,
those those two things, reallyproduct mindset and
communication skills are, Ithink, the major thing and if
(28:49):
you were to plot on a graph,your kind of evolution of that
buzzworthy sort of roles andtitles from your SIS admin, to
your infrastructure to yourdoubts in the near future, apart
from engineer, you know, if youhad time on the x axis and sort
of like communication on the yaxis, then you know, they would
(29:10):
plot on to the right, andthat's, that's, I think, kind of
how you want to think about it,right? Like platform engineer is
basically a DevOps engineer thatlistens and communicates better.
Eveline Oehrlich (29:21):
Beautiful. Wow, that sounds like a really
exciting career choice to makeif anybody out there wants to
know more, there is plenty moreto get involved in. I just say
that there is a fantastic reportand I think you will look about
the author of it. It's calledthe state of platform
engineering. So if anybody isinterested in then just look for
(29:43):
Luca Galanti and type in stateof platform engineering, because
there is all these differentways of getting involved in
Slack channels and communitiesand events. Super. This was
extremely insightful. I think.Now Now comes the fun part. So
besides doing these greatthought leadership pieces, doing
(30:05):
your work, talking to peoplelike me and platform
engineering, what do you do forfun Luca?
Luca Galante (30:13):
Um, well, so as I mentioned at the beginning, I'm
currently on Tenerife. And soit's been really fun over here.
I've been living out of eventactually, for the last couple of
weeks and surfing mostly, reallyearly in the morning, because
we're one hour behind here. Butthat's, I think, one of the
(30:33):
things that I do for fun, youknow, just just being in the
water being in nature, that's,that's really good times for me.
And the other one, frankly, as agood Italian is just eating and
drinking good wine. So, youknow, simple, simple person
simple needs, but does go prettyfar for me.
Eveline Oehrlich (30:51):
Fantastic. We have to meet in person, I have a
van. I love Italian food. Oh,amazing. I cannot serve you with
so I do like the water. So maybewe get to meet one time
somewhere in a place. MaybeJennifer, this has been
fantastic. Thank you. We've beentalking to Luke. I like Galanti
product owner, product master,and really a great thought
(31:15):
leader on platform engineering.Luca, thank you again so much
for joining me today on humansof DevOps podcast.
Luca Galante (31:23):
Thank you so much for having me. And if they can,
if I may, just because youmentioned the the events. If
people want to learn more aboutplatform engineering, I would
encourage them to check outplatform eg.org. And that's
really the home of the communityand platform khan.com. That's
the event that's coming up inJune this year. It's free,
(31:45):
anybody can sign up and we'llhave the best of the best really
of DevOps and platformengineering thought leaders
there. So I invite everybody tojoin us over there.
Eveline Oehrlich (31:56):
Great, thank you, humans. Yes, humans of
DevOps podcast is produced byDevOps Institute. Our audio
production team includes Juliapape, Daniel Newman, Schultz and
Brendon lay. Those are the threefolks I want to have a shout out
to because they're wonderful. Iam humans of DevOps podcast,
(32:16):
executive producer Evelynearlyish. If you would like to
join us on a podcast, pleasecontact us at humans have dev
ops podcast at DevOpsinstitute.com. Do I sound like a
virtual agent? Maybe? Anyway,have a great day. I'm Emily,
talk to you soon.
Narrator (32:39):
Thanks for listening to this episode of the humans of
DevOps podcast. Don't forget tojoin our global community to get
access to even more greatresources like this. Until next
time, remember, you are part ofsomething bigger than yourself.
You belong
Popular Podcasts
Dateline NBC
Current and classic episodes, featuring compelling true-crime mysteries, powerful documentaries and in-depth investigations. Follow now to get the latest episodes of Dateline NBC completely free, or subscribe to Dateline Premium for ad-free listening and exclusive bonus content: DatelinePremium.com
24/7 News: The Latest
The latest news in 4 minutes updated every hour, every day.
Therapy Gecko
An unlicensed lizard psychologist travels the universe talking to strangers about absolutely nothing. TO CALL THE GECKO: follow me on https://www.twitch.tv/lyleforever to get a notification for when I am taking calls. I am usually live Mondays, Wednesdays, and Fridays but lately a lot of other times too. I am a gecko.