Is DevOps still a thing?

December 21, 2022

Season 1, Episode 3

Is there a place for DevOps in a Cloud Native world? And what does Cloud Native really mean? Appvia founders Jon Shanks and Jay Keshur take on these questions and more. Come for the answers. Stay for the laughs. You don’t want to miss this one!

In This Episode, You Will Learn:

  • The fundamentals of the Spotify model and its application to cloud operations.
  • The significance of squads, chapters, and agile delivery teams in the modern tech landscape.
  • The balance between autonomy and the need for standards in a decentralized structure.
  • The potential challenges and inefficiencies that can arise without a centralized approach.

Themes Covered in the Podcast:

  1. The Spotify Model: An exploration of Spotify’s unique approach to team organization and its relevance to cloud operations.
  2. Squads and Chapters: A deep dive into the structure of squads and chapters, and how they differ from traditional team setups.
  3. Agile Delivery Teams: The importance of agile methodologies in today’s fast-paced tech world.
  4. Centralization vs. Autonomy: The pros and cons of having a centralized approach versus giving teams full autonomy.
  5. The Role of Platforms: How platforms can simplify processes for developers and ensure consistency across teams.

Quick Takeaways:

  1. Squad: A multidisciplinary delivery team in the Spotify model.
  2. Chapter: Refers to different skills or functions within the Spotify model.
  3. Tribe: A collection of squads and chapters, similar to a business unit.
  4. Guild: Groups with common interests, akin to centers of excellence.
  5. Backstage: A tool developed to simplify processes for developers.
  6. Autonomy: The freedom given to teams to operate independently.
  7. Agile: A set of methodologies aimed at improving software development processes.
  8. Cloud Operations: The management and delivery of cloud services.
  9. Centralization: A centralized approach to managing teams and resources.
  10. Operating Model: The way an organization functions and delivers projects.

Follow for more:
Jon Shanks: LinkedIn
Jay Keshur: LinkedIn
Jon & Jay’s startup: Appvia


0:03so I’m here to help your business some of the devsec finops cloud native institution of Jacob show Services

0:12I mean it’s now called Cash Ops you know for two thousand pounds

0:24hello and welcome to Cloud unplugged this is season two episode five I’m John

0:30Shanks and I’m Jacob Shaw um so we put the world to rights in the last episode uh feel free to have a

0:37listen apparently I got the most passionate I’ve ever been yeah

0:43um so yeah last time we spoke about um

0:48kind of just the how to make decisions so that you don’t have high cost or insecure

0:54um I guess Cloud delivery and whether you can stay lean

1:00um today we’re going to talk about like Cloud native what does cloud native mean devops what does devops mean to Cloud

1:06native um obviously they’re separate things but kind of overlay a little bit on

1:11um Quality and expectations around applications um and then portability and the other

1:19advantages and things like that and whether it’s important whether you should do it whether you shouldn’t do it so

1:24I guess Cloud native J sure as always I’ll ask you questions and then

1:30violently disagree with everything yeah I might ask you a few questions here because uh you were ahead of devops a

1:35few times yeah so I’m going to ask you what devops yeah I know you love it uh

1:42well before back in the day yeah um how people were delivering was very

1:48siled um network teams everybody we talked about before having big spreadsheets and

1:54releases and lots of the stuff um and devops was a way for operational

1:59aspects of the delivery to sit close to the application and the application team become familiar with those operational

2:07aspects and to improve the application delivery um speed and you know time to Market Etc

2:15and then the quality of what was being produced as well so that’s where it originated from

2:21um and there’s been a lot of improvements on quality but then obviously the industry has also evolved so it was a methodology obviously but it

2:30was trying to solve a very specific problem back then and that problem probably does still exist I’m not saying

2:37it’s gone away there are people that do have lots of different teams that maybe are delivering in kind of slow traditional ways and

2:45maybe you’re looking at just going to Agile for the first time and looking at devops and things like that so obviously

2:50depends on where you are as business but there are also a bunch of other companies that have kind of moved to

2:56cloud and you know there’s things like Cloud native and microservice architectures and all these other things

3:05um and the agile ways of working have already given people quite a quicker way of track tracking velocity yeah you know

3:13story pointing um especially using scrum Etc so there are delivery methodologies that are

3:20around devops too um that don’t measure quality right so that’s just about speed but the

3:26definition of done you could argue could measure those things exactly right so you could shift that stuff at a two

3:32different place set some requirements on what the definition of done means um Define the software delivery life

3:39cycle and you know around all of your delivery and kind of Encompass it as one

3:45um I don’t think it was supposed to be a permanent role is it supposed to be a cultural change in some ways I think are

3:52you saying you don’t like the devops engineering role that has now existed I don’t think it really means anything

3:59really I think I think it’s become it’s a category of somebody’s CV but the

4:06CVS will look different so at devops engineer for x and a devops engineer for

4:14y you know what goes in them most of the time is not the methodologies yeah and

4:20not so much necessarily the responsibilities towards the business outcome but is now more about the

4:25Technologies you understand totally so yeah in that sense I don’t really agree

4:31with it necessarily but I think you’ve just kind of got used to it so it doesn’t it kind of means what it means

4:36you just missing you just reinterpret the meaning which is what I’ve done which is like oh it’s a is somebody

4:42helping with application delivery that’s kind of like got some skills so let’s say you’ve got the uh the chance to like

4:50coin a new term right you can you can create a brand new term for this engine

4:55for this person that is delivering on this business outcome for this organization what would you call it

5:02well I guess it depends there are problems no because there are probably terms already yeah that exist they would

5:10just that do get used but just less mainstream like a cloud engineer would

5:15make sense yeah right because you’re engineering for the cloud that’s your job um you might Encompass understanding

5:22devops right as part of that but it doesn’t mean that they are a devops yeah if that makes sense and you might ask

5:28questions in the interview process where they understand it SRE again another role very specific very much not devops

5:35again what is that SRE site where liability engineering basically always been responsible for the scale and

5:41reliability of a service um and being quite embedded sometimes in specific services and improving those

5:47services in terms of reliability and scale right um and if you’re a large product

5:53business and you’re seeing that as a challenge and you feel that it does require certain expertise who does

5:58understand reliability really well and kind of think like that and puts the Solutions in place to measure things well

6:04then that makes perfect sense um so automation Engineers sometimes

6:10there’s those um but I mean there’s lots of different terms that you could choose from

6:17that isn’t necessarily devops engineer either yeah it’s true I mean automation engineer I’ve seen that um kind of

6:23banded around a lot and it probably yeah it’s so similar right because you are automating all the things but it could

6:30be anything from infrastructure to tests to applications to like automating your

6:36coffee in the morning like yeah it could absolutely mean absolutely anything grommet in yeah

6:43that’s how I get that’s how I get changed in the morning yeah I just pajamas automated come off try automated

6:49jeans and t-shirt I wear the same thing every day so yeah um yeah so okay so you’ve got

6:56um devops which is maybe like a cloud engineer SRE uh automation person

7:03um there’s a couple of other things that you touched on there which was um Cloud native um so what is what does cloud native

7:09mean to you what does it mean to me what does it mean to you and then let me tell you

7:14what it means to me all right this is my cloud native means to me so everyone

7:19listens clearly listen up yeah listen up listen clearly

7:24it’s a way of offsetting uh offsetting complexity outside of your

7:31application to simplify it to reduce um I guess improve on the efficiency of

7:37that so it’s giving the flexibility portability um I guess scalability reliability of

7:44services and to capitalize on those things that are Cloud as in things that become native to the cloud you can use

7:51all of those to your advantage if you do that means you’ve shrunk how much your application has to worry about right or

7:58other ancillary things to the application that might have needed to be there or your business right or your business overall yeah so like a database

8:05a message queue all those things you don’t have to bundle it up as your app you don’t have to necessarily run those

8:11Services there are cloud services there for you to consume yeah um and they can be highly scalable and

8:17reliable services that then improves your service because obviously you’re consuming something that’s reliable Etc so there’s good knock-on effects and

8:24principles around that um and kind of microservice in design usually not necessarily so but most of

8:32the time you’d kind of similar right um I guess architecture is is a big part

8:37of being sort of cloud native yeah um and there are more sort of common

8:43patterns of say architecture nowadays you know you used to go from sort of client server models to

8:49um now you’ve got you know the kind of microservice architecture function as a service serverless all of that stuff

8:57um like where I guess what do you think is right for a business that the types

9:03of business that we talk about um so the you know medium mid-sized Enterprises or Enterprises delivering

9:10software is is there any one right answer is it is it custom to what they’re building well

9:17I mean there’s different schools of thought I suppose here so there’s a lot of services in the cloud already

9:25um that you could choose from which you’ll yield a lot of advantages serverless be one of them not everything

9:31needs constant run time not everything’s processing Everything 100 of the time they’re probably advantages to

9:37event-based architectures right do this thing when this event comes in

9:42um now some of those things and this is where it gets a bit complicated are very

9:47bespoke to One Cloud um some things in the industry are more agnostic to other clouds

9:54um and but then also some of those things that are agnostic are also very complicated so there’s a bit of a you

10:01know kubernetes being one of them let’s be honest it’s a big fast ecosystem however yields lots of advantages but

10:07also inherently quite complicated um and isn’t that simple it might be

10:12simple for some people but it wouldn’t be simple for somebody that’s new to it yeah right who’s got to learn all these new Concepts and and but at the same

10:18time there’s lots of products there that simplify it too so you could don’t have to necessarily know the ins and outs of

10:25all of it um and then I guess that gives you multi-cloud stuff but if you you depends

10:33on your definition of multi-cloud if I want to Port things around easily but I’m kind of very customized into one that’s harder

10:40um at the same time the Frameworks like Sam for serverless maybe they’re a bit more agnostic maybe

10:46it doesn’t take as much to Port things over so I think from my experience and this is going

10:52down a rabbit hole when it depends um what you need to keep your eye on is

10:57infrastructure and networks because usually the things that become difficult

11:03to maintain or pour or move between clouds are those things is very rare the app the app is a library

11:10that’s consuming some service like S3 bucket RDS whatever it doesn’t care necessarily too much

11:18right I’m in a container great I mean I don’t care yeah well you know what scheduling yeah who cares yeah right

11:23kubernetes is a manifest can be that could be something else but changing the libraries quite quick

11:29for the team because usually that team’s iterating on bugs anyway and doing stuff so they can easily move to a like new

11:35library and test the things that are usually complicated aren’t those things it’s usually how the

11:41cloud has been constructed underneath the team how it’s operating beneath the

11:47team that’s delivering that then just can’t be moved very easily oh we can’t really move this to another Cloud vendor

11:54right it’s going to be like eight months and we’ve got to like repeat all this other stuff and then it’ll need security stuff and then we need to get it pen

12:00tested and they’re like there’s all these things you just think well wow that’s not very Cloud native exactly

12:06it’s almost the costume the cost of the business is almost higher in terms like the cost of development team tiny

12:13because they’re only they only care about you know the unit test integration test of their service and maybe wrapping

12:19a library or whatever yeah that is using a service but I guess the overall cost of the business to change it could be

12:26huge right to make it multi-cloud or whatever else and which kind of defeats the purpose of the cloud it’s Cloud

12:32native for me Cloud native is supposed to be like the flexibility in the portability but if your Downstream

12:39infrastructure implementation detail is preventing that possibility flexibility then something’s gone wrong with your

12:45Cloud native implementation yeah you know like you’re not particularly Cloud native you might be from an application

12:52perspective but not from an enablement perspective you’re in Cloud you’re in a cloud native Cloud native if it’s that

12:59hard for things and you know so I guess to answer your question on like what do you do in the

13:06services and what does it all mean and the answer I think you’ve just got to keep your eye on how you’re going about it more than

13:13anything else because that’s the thing that dictates the total cost in the end or

13:18the lack of flexibility I’d say and it’s a bit of uh you know it’s I guess it’s a

13:24um you have to weigh it up because you can be totally Cloud native and rely on

13:30all of the um sort of cloud SAS Solutions but then

13:35you might be locked into that cloud vendor right because which might be fine too which might be fine yeah exactly that’s not it’s not that’s what I’m

13:41saying it just just depends on on the organization on the organization yeah exactly and there were a lot of like

13:49you know stitching bits and Bobs together even in the cloud you kind of got to Stitch things right nothing

13:54nothing just works out the box and gives you absolutely everything you need straight away yeah right so you know

14:01you’ve got to turn things on you’ve got to enable the apis right that to use it’s like you’ve got to configure them to enable them properly it’s like so

14:08wear a cloud watch do I have logs yeah if you want yeah you can get them but do

14:14you want them how much is it going to cost you exactly yeah how do you want them where do you want them exactly so

14:19just one thing requires like questions and thought and then you’re like well do

14:25I want all that title Cloud vendor necessarily I don’t know maybe it doesn’t maybe that’s fine you can use it

14:30maybe it isn’t fine maybe you’re best off going to Elk or something buying their SAS right but then it’s a listing across all the clouds exactly so there

14:37might be better Alternatives that aren’t the cloud vendor specifically that give you a bit more agnosticity and

14:43portability yeah for certain Services again though you don’t need to be an elk

14:48team right pretending to be a SAS business inside of another business providing a service back to all the

14:55teams like you are us internal SAS company because we’ve been down that path by accidentally yeah do you know

15:02you go down that path and then it’s impossible because it’s like you’re trying to scale it and you’re now acting

15:08like a SAS business trying to run scale um on a service that isn’t even what you

15:13created exactly you’re just running it and you can your consumers I guess the you know application developers that are

15:20just trying to iterate on their services are like I don’t just give me some bloody logs I don’t I it’s so expensive

15:27to get them this this cost that you’re presenting to me because you’ve decided that you want to run your own elk stack

15:34or something locally you know um locally to the business is is expensive and just really expensive yeah um and and I guess

15:41people fall into that same pattern or they have fallen into the same pattern uh even with kubernetes right so you

15:49know back before the um the cloud vendors had their own um managed uh kubernetes services

15:57you spend so much time managing upgrading patching all of these services

16:03that you can now kind of just get commodity off the shelf um where it’s even taking care of the

16:09master plan you don’t see those and actually the newer Services where you’ve

16:14got things like fargate and where you don’t even see the the worker nodes

16:21um so is I guess when we’re bringing it back to sort of cloud native is there is it a gradient you know how

16:30what is some things is it Cloud native and it is or it isn’t or is there a bit

16:37of a kind of sliding scale of something being more Cloud native than less

16:43I mean I hope you’re not suggesting that someone goes off to create an equivalent to Dora metrics for cloud native

16:50specific things to measure how Cloud native you are um yeah I mean

16:57probably is a sliding scale it must be but I don’t I don’t think that’s matters necessarily It’s Tricky though to know I

17:06think what you have to bear in mind is how much it’s helped the business and

17:13you need some contextual stuff there to kind of ground it in it’s like if you’ve

17:18managed to write an app quickly and consume a lot of the services that the car vendor gave you and it

17:23functioned really well and there’s really good slas for those things yeah right and now you’ve got good agreements

17:29in place and good support agreements and running that stuff you know the alternative would have been well I would like to run all that stuff yeah yeah and

17:35it would have cost me all this money then obviously doesn’t make sense exactly um and so I think if the application is

17:41designed in a way that’s taken advantage of everything that’s there that’s all it is it’s like capitalizing on value at an

17:47application Level um and our team level yeah but I think it is more on the

17:53infrastructure side still tends to lend itself to the burden but you’re right things are improving there

18:00um but it’s a funny Paradigm because we’re talking about like teams and Central teams

18:05but there’s a juxtaposition between those things which is the central team isn’t the business unit

18:12yeah and the business unit owns the users a lot of the time on either end

18:18and potentially the data depends how you’ve organized yourself but if that out if there was an outage

18:24planned or if there was a service maintenance window or something they would have to comms to their users

18:31to let them know yeah right because maybe they don’t have to let them know but yeah potentially but those users

18:36have an expectation on their service yes right and so the minute those users have an expectation on their service and when

18:42a very consumer driven world right people get very frustrated there’s probably someone else I can go right so it’s like oh well they had an outage for

18:49doing it they didn’t even let me know I’m going to use you know whoever else right because it’s competitive industry

18:54so that expectation has to be met and another team I mean this is why I

19:01think or like Auto upgrades is just should be a thing you know we should be shifting left to

19:07the business units I don’t think this kind of Dev shift in left I think these are elements have shifted left but there’s also shifting left operationally

19:13to the business unit for cloud because they own the risk operationally in the end so how can they be in control for

19:20themselves on managing the operational risk yeah to the point of not needing to be an expert

19:26that’s the kind of dichotomy um but at the same time I don’t I don’t

19:32have a regional person knocking door-to-door who upgrades my Mac for me right there’s a button that comes on

19:38it’s like do you want to upgrade now or do you want to do it like I mean that was a solved problem it’s not like oh God that’s the first language the person

19:44that came to my house and asked her anyway oh my God that was that a scam you were drunk

19:52just handed my phone overgrade me all your data from doing

19:57that upgrade cheese um yeah so I I think you know the funny

20:04thing with our space and we talk about devops and things is other Industries do a really good job on user experience

20:11right they’ve thought that through you know the es technology is everywhere that doesn’t mean experience has to be

20:18really highly complicated so you can simplify things you can make it easier

20:23so you can shift these things left to business units that’s definitely possible yeah

20:28um so yeah I mean you can make Cloud native easier and simpler and also

20:35portable and also flexible and also multi-cloud that is also possible and you can actually make it easier for the

20:41teams to not even be affected by so you can shift it left and you can make it so

20:47simple that they’re not even affected by you know infrastructure behind the

20:52scenes being upgraded or whatever like all of those things again we were talking about you know making like the

21:00right thing the easiest thing to do and or or the only thing that you’re able to

21:06do and if you’ve made those choices and and made that uh available to the users

21:13of your your service or whatever that you’re providing to to the application developers then you’re going to be in a

21:21better position their users are going to be in a better position yeah um because your users are theirs because your users

21:27are them them and they have users and then they have users down straight exactly so your internal customer in

21:33this case is the downstream development teams if you are operating that kind of central capacity I guess yeah otherwise

21:39the team could hire you and you shouldn’t be responsible for the team but even then you’ve been hired to enable that team either way like you

21:46know you’re that you have your own users how you’re serving those users you know does need thought right you

21:53can’t just command with a bag of your tools and Jeremy just land in and be

21:59like well this is just what we’re going to use this is what I’ve always used exactly you know and it’s like well I don’t know like there’s a better ways

22:05it’s 2022 surely we’ve I feel like it always goes on way it’s it’s uh this is

22:10what we’re going to use because this is the only way that I know how to do it slash this is what I’ve always used or

22:16this is what we’re going to use because someone told me it’s the next best thing and you know I’ve got no experience of

22:22it I’ve got no experience of this business and the requirements that I’m trying to meet but it’s uh the next best

22:28thing low code you know let’s let’s do it yeah just technology is the right one yeah some technology let’s just sorry

22:34let’s just do it on Twitter yeah I need it on my CV I need it on my CV yeah it’s

22:40definitely definitely going to be CV LED 100 stuff I’d like to think though people weren’t motivated like that I

22:45don’t I don’t know I mean maybe I have a bit of a half a glass

22:50half full type mentality but I just think the industry if you’re into

22:55technology then you’re obviously going to find it stimulating and you’re obviously going to enjoy the fact that

23:01new things are coming out and you really enjoy it and you love the research and love trying things and tinkering nothing

23:06wrong with that um it’s good it means you’re exploring you totally but you need to also apply it to the business in

23:12the right way and so it can’t just be that I don’t I think that’s the disconnect I don’t think it’s that people just want it on their CV I mean

23:18maybe there are people out there that do want it on the CV I don’t know but I think maybe they do think it’s maybe

23:25right but they don’t know if it is or it isn’t because they never used it themselves just just measurement right it always leads back for me it always

23:32leads back to to measuring whether or not because if you if you think about it

23:37you know engineering it’s quite scientific you know you do something

23:43um and it and sorry it can be quite scientific in that you have a hypothesis

23:48that this change that you’re going to make gonna make has an impact but you don’t

23:55actually people don’t necessarily think of it in that way and they’re not measuring it before and after to figure

24:02out whether or not their hypothesis and their you know how they’ve experimented and implemented their thing has actually

24:10worked yeah like how do you how do you know um and when it’s user-facing

24:18um you know to a business user application developers are are much better at kind of doing this because

24:24they might have a bit more of a connection to say the revenue that their website is driving or the click-through

24:32rate of of um you know hotels being booked or something like that whereas

24:38um the kind of that that Central team platform team

24:43Cole sorry about that uh everybody who had a run out of battery on the very kind of classic

24:50um Jay was the devops engineer on this guy

24:56so there’s um some of the user requirements weren’t gathered properly on some of this stuff

25:02um yeah sorry about that we just had a bit of a battery outage on something I had to like

25:08basically swap batteries cool story John thanks um so where were we we were talking about

25:14um the teams having and users of their own and meeting those requirements and then

25:20building in line with that and making them as efficient as possible and then you were

25:25talking about like measurements basically obviously tracking that did that actually bring the value exactly

25:31that before and after I I made a change hypothesis you were talking about yeah I made a change that change cost me some

25:38time um and uh it cost the business some amount of money

25:44um and does that is that change been beneficial or not has it meant that

25:50people on my platform or whatever can now iterate more quickly um does it mean that they there’s less

25:57security flaws all of all of that stuff I just don’t think that it gets measured in that way really

26:03in in this uh in this environment in this kind of world that we’re in yeah

26:08because they’re not business measurements yeah exactly and I’m not sure I’ve yet to have seen

26:15um they’re probably Downstream like you were saying before with devops it was Downstream it’s like was that did that

26:22code get deployed faster that’s it right it’s not it but that’s not specific about like it’s almost a representation

26:29of the downstream customer and so it kind of makes sense but like it could have cost loads of money to do

26:35that it’s not all encompassing it’s not all-encompassing like maybe the effort involved for that result was way off

26:42yeah right it could cost you could have cost you you know you could do anything and be like oh by the way how much did that cost you cost you three million

26:49pounds great okay so I’m going to tell you I could have done that for two thousand pounds what exactly oh my God

26:55you realize Jesus so like there’s no context give this guy some money two thousand pounds two thousand pounds just

27:02two thousand pounds geez two thousand pounds to UJ right now I can I can do

27:07this where’s my checkbook yeah where is my check exactly get it

27:13um but yeah so you’re right I don’t think there is structure around teams in the right way

27:19to know what the real business outcomes are Beyond just standard delivery ones

27:26um and therefore there isn’t a true measure of like the actual value in the end to the business in the right way and

27:32then things were kind of ranting about before is like then you’ve got to go and buy something because that TV oh yeah well you know we

27:39don’t know yeah exactly um you could always like come in to businesses and just say all the words

27:45yeah couldn’t you you could say like look I’m a devops club native specialist who secops who yeah who likes to do a

27:53bit of SEC so we call it deaf second I do take finances very seriously so I do

27:59understand hops right yes yeah um so I’m here to help your business some of the devsec finops cloud native

28:05institution of Jacob Services I mean it’s now called Cash Ops you know

28:13it’s just casual you know for two thousand pounds all of these services

28:19um you know but we do do policy as code and it is very githubs and infrastructure is co-centric and we do

28:25use continuous integration continuously my name’s not Chris did you did you think it was okay these are all the things we do yeah but it’s a bit of a

28:32funny space hard to uh all the words yeah all the words that make no sense in

28:37some ways in the end um but make a lot of sense individually about what they mean but so I guess

28:44Cloud native don’t know where we obviously went off on a massive tangent about all this I’ve got to go off on

28:49that um Cloud native um there’s the value of it the value of

28:55it there’s like you know um principles that you might kind of adopt while doing

29:00Cloud native how much responsibility you’re offloading to the cloud versus

29:06how much you’re taking internally how much you’re offloading to a service that you might procure products Etc versus

29:13doing it internally and then what principles you might want to push down onto the application teams delivering in

29:21a cloud-native way right and that’s one thing that we haven’t touched about touched upon yet um obviously you know in previous jobs

29:28you’ve helped write sort of technical standard standards for teams and some of

29:34this is like duplicated into our Factor applications now so if you want to talk about what that means

29:40uh what the technical standards mean just just um like 12 Factor applications what are they you know or what what

29:48principles might you have um that you’ve set sort of as as a

29:54center of excellence type type approach that you that are good for application

29:59developers that are developing in Cloud uh I mean that’s massive in some ways but I mean yeah it’s a big topic but I

30:05guess you know everything has to boil down to again an outcome you’re striving

30:11for so um there are like you know you want that

30:18language if you choose certain languages you want the ancillary things to be as

30:24templatable as possible right so I guess that’s the thing like logging standards create a library reuse it have the

30:30standards baked into your languages um the aim is to just reduce as much

30:36complexity as possible in the application um and then to also offset some of those

30:42Services independently so like we talked about before you talk about payment stuff yeah um it doesn’t matter what the service is

30:48but that service in theory should be able to have its own release cycle and also scale independently

30:55um there are complexities with microservices right keeping track of everything everything moving around you

31:01how do you test that everything’s going to work version it exactly so backwards compatibility comes in there and setting

31:07those standards that you know I think Google’s one is if somebody’s consumed this API at that point I’ve got a

31:13contract to support that user um right they’re now using it as an expectation on it

31:18so as long as you kind of have those principles laid out for yourselves then you know you can just work out good ways

31:24of engineering at the application Level to always be backwards compatible version your apis properly as you’re

31:31moving maybe you can then get the data to say well no one’s using the old API anymore now we can decommission that

31:36logic outside we don’t have to have that version so there’s many best practices

31:43um but if you’re doing microservices you need those rules I think in place it’s just to make it easier for yourselves

31:49and to have different teams yeah um otherwise you’re kind of like a monolithic team under the guise of a

31:55micro service just because you can also I guess there’s also like organizational ways of achieving something like that

32:01right like you could go off and like have an API Gateway and start just

32:07pushing keep your old uh version there and just start building out your new one

32:13iterated you know so like the how um your the onus on the application

32:20um development team changes depending on how you’ve enabled them

32:25um in the right ways yeah I’m not a fan of that just because of hot fixes and things like that he’s still you know if

32:32there’s something wrong with that version you’ve still got to go and fix it if you’ve got multiple places exactly

32:38so I think your best of manager in code personally where you have different version M Points yeah inside your code

32:44and that’s very easily tangible and you can see it yeah right and you’ve organized your code in a really good way and effective way yeah but obviously

32:50there’s many ways it’s gonna have to take it that way but um yeah and then just basically

32:55obviously lots of tests you know failing fast those principles mocking things

33:01stubbing things um there’s even cloud services as local stack there’s stuff like that you can

33:07use right that you can even have it local on your machine you can test that these things work yeah um even when you’re using Cloud mocked

33:14cloud services um so there are many ways to do that and the testing period with the triangle

33:20thing obviously lots of unit tests and very small amounts of end to end at the top that’s what you’re aiming for yeah

33:27um so yeah so they’re all the kind of like standards obviously there’s zillions of them

33:33um way more than the ones I’ve listed but they the main things on quality benchmarks nice

33:39what about what would you said if I missed something obvious uh no I think uh that’s that’s pretty much it um

33:45there’s obviously the ways to kind of get your um your application running so splitting

33:52out kind of um environment config in a way that makes sense yeah using environment

33:57variables yeah exactly yeah application yeah exactly so just just things that mean that you can scale well like build

34:05your application once and run it in the most efficient way by letting it sort of

34:11read config from something that’s easily um sort of insertable as possible and

34:17then managing um the release cycles of everything well

34:23um uh remove it like in increasing the

34:29um automation around your your products the um the quality of your products independently

34:34um and having a um what’s it called uh

34:41an architecture that is decoupled um essentially that’s that’s all it boils down to yeah because some of them

34:47even with the with the decoupled architectures some things are people things some

34:52people can be technology things and I remember people talking about oh I need to release this entire all my dependencies

34:59to test everything works and I was like well you kind of don’t right you could just say actually you as a team response

35:06for this microservice is responsible for managing and supporting it for all environments yeah right so actually

35:12there’s an expectation on another team that will be a Dev version of this on the latest version of the thing which I

35:18also can check the version of and then I can test whether I’m releasing when I’m released to Dev that everything works according to that version thing and

35:25you’ve got your life cycle it’s like it doesn’t have to be that I’m now responsible for losing your Dev thing yeah to see if it works because like I

35:32don’t even know it works necessarily right I don’t know how to you know so it seems a bit odd when people move it it’s

35:38like operate it as if you um like the teams presenting Services yeah because

35:43someone’s got to support that service eventually yeah so you’ve got a contract you’ve got a both a software contract

35:50and a contract to the rest of your business yeah on everything that you do um there are

35:56many different ways to think about the problem but the instinctive way is they’ll want to solve it all

36:01by releasing everything as once in their own environments and then testing everything works and like well that’s possible but you could be spending ages

36:07trying to get someone else’s thing to actually work yeah um which isn’t a great use of your time to find out that

36:14actually it worked once you’ve got it posted those unnecessary times yeah but

36:20not business value but yeah um cool well we had a little bit of a hiccup in this episode

36:26um probably like might do some sneaky edits yeah exactly to see what we can do skills to the max obviously as always

36:34what I hear from you any topics of Interest please reach out if you don’t agree with us please reach out and

36:40um or if you do agree with us reach out and have a conversation about it um but yes we shall speak to you all

36:47next time and again we’ll just pick a random topic who knows what we’re going to be speaking about

36:53um I don’t you do do you yeah I don’t have a clue anymore um so yeah and stay tuned next week when

37:01we’ll release another episode great thanks mate bye [Music]

37:16thank you

English (auto-generated)AllFrom Cloud UnpluggedDevOpsRelatedListenableRecently uploaded