0:00

Netlify is a popular hosting platform

0:02

that provides build, deploy, and serverless

0:04

backend services for web apps. The

0:07

platform enables deployment directly from source

0:09

files stored in a version control

0:11

system like GitHub. Erika

0:13

Passani is a senior software engineer

0:16

at Netlify. She joins

0:18

the show to talk about how she got started

0:20

at Netlify, edge computing, edge functions,

0:23

private integrations, and more. This

0:26

episode is hosted by Josh Goldberg,

0:28

an independent full-time open source developer.

0:31

Josh works on projects in the

0:33

TypeScript ecosystem, most notably TypeScript ES

0:35

Slint, the tooling that enables ES

0:37

Slint and prettier to run on

0:39

TypeScript code. Josh is also

0:42

the author of the O'Reilly Learning TypeScript book,

0:45

a Microsoft MVP for developer

0:47

technologies and a live code

0:49

streamer on Twitch. Find

0:51

Josh on BlueSky, Mastodon,

0:53

Twitter, Twitch, YouTube, and.com

0:55

as Joshua K. Goldberg.

1:10

With me today is Erika Passani, senior

1:12

software engineer at Netlify. Erika, welcome to

1:15

Software Engineering Daily. How's it going? It's

1:18

going great. Thank you so much for having me

1:20

on today. It's my pleasure. I have to note,

1:22

we'll talk about this at the end of the

1:24

interview, but I'm on your About page on your

1:26

website, and there is significantly more space dedicated to

1:28

your dog, Ada, than anything else on that page.

1:31

That's great. Thank you. Thank

1:33

you. Yeah, I feel like of Ada and

1:35

I, Ada is obviously the one people want to see when

1:37

they visit my website, so I made sure to feature her

1:39

very prominently as much as I could. She's

1:42

doing great on there. Props to the dog. But

1:45

let's talk about you for a little bit. You're a software developer.

1:47

How did that come to be? Honestly, it

1:49

was a little bit of a roundabout journey. I

1:51

was definitely that kid growing up that

1:54

liked hearing the clickety-clack of keyboards, and

1:56

I loved playing video games. And so

1:58

at one point... early on, I remember

2:00

thinking, yeah, I'm going to be like a lot of

2:02

developers. I imagine, yeah, I'm going to work on video

2:05

games. This is going to be great. And

2:07

then somewhere around high school, I ended up

2:10

thinking that I think I took like a general

2:12

tech course in high school, did a bit of

2:14

programming. It felt like it wasn't for me. So

2:17

I ended up deciding to not continue

2:19

with it and instead decided to

2:21

focus on history because I just love

2:23

that subject and a

2:26

bit of time later, it was like maybe two years

2:28

after I, this would have been like the 11th grade.

2:30

So I don't know how, how will that translate to

2:32

16, 17. I

2:35

was trying to get courses

2:37

to work out my schedule to work out. And

2:39

there was a bit of a clash and I

2:41

needed to pick a course to fill this last

2:43

slot and it was either hair and

2:46

makeup styling, which is if anyone knows me for

2:48

more than 10 minutes, you know, that is definitely

2:50

not my jam or retaking

2:52

programming. And I had a bunch

2:54

of friends that were in that class. So I was like,

2:56

sure. I used to dream of

2:59

maybe being a video game programmer. I'll

3:01

take the class again. I'm sure it'll

3:03

be fine. And around

3:05

that time, my mom was actually in the

3:07

middle of a career change. She was going back

3:09

to school to be a teacher and she

3:12

was going through teachers colleges to that final

3:14

year in Ontario where you're, you're doing placements

3:16

and stuff. And she was like, unless you

3:19

are really set on like,

3:21

you love history so much and you really like

3:23

the idea of teaching, this might not be for

3:25

you. So you might want to consider something

3:27

else. Cause she was just having a really hard time with

3:29

it. And I'd never pictured myself

3:32

being a math whiz. Like that's kind of the

3:34

picture I grew up. I guess those software engineers

3:36

is you had to be really good at math.

3:38

You had to be good at snacks. You had

3:40

to be living and breathing code throughout

3:43

high school. And then you, you go into

3:45

college or university continuing to

3:47

program. And fortunately

3:50

for me, and like, honestly, half of my career

3:52

is just a lucky set of coincidences where I

3:54

spoke to the right person at the right time.

3:56

But I had a classmate that actually was like,

3:59

Eric, you're actually. pretty decent at this,

4:01

you should consider studying it and pursuing it

4:03

as a career. And if you don't like

4:05

it, you can always go back to studying

4:07

history. So that's what I did. I ended

4:10

up like going into programming and university. And

4:12

then there were a few times I almost

4:14

dropped out, like in my second year of

4:17

university, I almost quit and switched things because

4:19

I found it not my

4:21

style. And again, I

4:23

ended up that so in Canada, we have this

4:25

concept of co ops, I think they're paid internships

4:28

in the US. And I

4:30

had this fantastic co op with this amazing team.

4:32

And I just they kept me sticking with them.

4:34

And I had these incredible mentors. And then, yeah,

4:37

just one after the next, I ended

4:39

up in these amazing jobs with amazing

4:41

mentors. And they kept pushing me and

4:43

helping me be better and got me

4:45

excited about tech and kind

4:48

of stuck with me. And here I am

4:50

today still describing myself as maybe not the

4:53

typical programmer because I identify more as a

4:55

humanities kid than a science kid. But

4:58

I still really love technology and all the things

5:00

that you can do with it. There

5:02

is something kind of funny about the large

5:04

number of people who achieve titles such as

5:06

senior software engineer and Netlify who are still

5:08

on the fence of whether to call themselves

5:10

a comp sci person. Was

5:12

there ever a moment in your career when

5:14

you really started to think more of yourself

5:17

as comp sci and humanities rather than just

5:19

pure humanities? Oh, that's a

5:21

good question. I think I can't

5:23

say a specific moment per se. But

5:25

I remember when I

5:28

became more aware of the

5:30

so called soft skills, actually

5:33

being very helpful and impactful in my

5:35

day to day job. And

5:37

especially like now fast forward to now where a

5:40

lot of us are working in a remote world,

5:42

things like being a good

5:44

written and verbal communicator, being

5:46

able to take these concepts and break it down

5:48

and change the language depending on who you're chatting

5:51

with, whether that's marketing

5:53

products, fellow engineers, even

5:56

like I would argue in this like my geeky

5:58

little theory, I think that one One of

6:00

the things I enjoyed about history was

6:02

seeing how all these various pieces tied

6:04

together into certain events. And I think

6:06

that that kind of mindset for me

6:08

really applies well in software engineering as

6:10

well. So when I started

6:12

realizing these kind of parallels, I

6:15

felt more comfortable identifying as

6:17

a software engineer instead of being more

6:19

exclusively like this humanities kid that

6:21

just managed to sneak in with the cool

6:23

software engineers. For sure.

6:26

It almost feels on the nose to compare

6:28

history to software because we do so much

6:30

archaeological digging within software. But I think you're

6:32

absolutely right that a lot of the core

6:34

skills of explaining, of researching, of learning why

6:37

are really applicable in day-to-day development. I

6:39

love that. And at least when I

6:41

was still visiting my old university, often

6:44

in talking with other students that were

6:46

looking for their paid internships and some

6:48

folks had maybe taken

6:51

computer science as a minor

6:53

or as a side thing and they were thinking

6:55

of transitioning and not feeling it was for them.

6:57

I think when I highlighted these parallels

7:00

between whatever humanities or even a

7:02

science that maybe wasn't, it

7:05

didn't seem directly related to computer science but

7:07

showing how it could be, I think got

7:09

more people excited about the potential of working

7:12

in technology, maybe not necessarily their full-time

7:14

careers, by being

7:16

aware of the possibilities that they can bring that

7:18

into their fields of study. Let's

7:27

face it, no one likes writing or

7:29

maintaining documentation. But when you start a

7:31

technical project or pick up a new

7:33

task, missing information can cost you valuable

7:36

time. GITbook is a technical knowledge platform

7:38

that fills that information gap, making it

7:40

easy for your team to capture, maintain,

7:42

and find information from a single source

7:44

of truth. Here's a couple of examples.

7:46

With GITsync, you can set up a

7:48

two-way sync between your repo and GITbook,

7:50

so you can turn markdown files into

7:52

awesome user-friendly docs. And if you make

7:54

a change in your code base or

7:56

a technical writer updates the docs in GITbook,

7:58

the edit sync between the two automatically.

8:00

Or what about when you need to find something

8:02

in your knowledge base? Forget about searching,

8:04

just ask Gitbook AI. You'll get a neat

8:07

summarized answer sourced directly from your docs. Those

8:09

are just a few examples of what Gitbook

8:11

can do, but it has plenty more tricks

8:13

up its sleeve. So why not give it

8:15

a try? Head to gitbook.com to find out

8:18

more. Do

8:26

you have any one particular piece of

8:28

advice that you would want to impart on anyone who

8:30

is like you many years ago now listening to

8:32

the podcast? Oh, I

8:35

would probably say like don't feel like

8:37

you need, this isn't gonna be like a very

8:40

succinct piece of advice, but don't feel like you need to

8:42

focus really hard on the latest

8:44

frameworks or the latest technologies and

8:46

instead keep an open mind and

8:48

stay curious and maybe follow your

8:51

instincts on what interests you because you honestly

8:53

will never know what thing

8:55

you're interested in, how that might

8:58

tie into technology and you will be that

9:00

person who is uniquely positioned to be able to

9:02

draw that relationship together and maybe be able to do something

9:05

no one else can do. I think

9:07

that's the best thing I can give is just stay curious, stay

9:10

just enjoy what you're looking into because you know

9:12

careers are long and tech

9:14

is changing so quickly and I think just being exposed

9:16

to so many things is only good for you. That's

9:19

absolutely true and I think it's doubly true

9:21

given your career history. Have you ever actually

9:23

gone into games because it looks like you've had

9:25

quite a few areas of tech but

9:28

everything except that original. Never games,

9:30

no. I briefly dabbled

9:32

with the idea as a

9:34

side project doing a, I

9:37

always loved the game Snake in

9:39

high school like on you know the older cell phones

9:41

that that's one of the few games that you could

9:43

have and I love the idea of building a little

9:45

Nintendo version of Snake where Yoshi's the head of the

9:47

snake and the eggs pop out and so that was my

9:49

only like idea of like oh I can make a game

9:52

but after hearing about

9:54

some of the challenges of working in the

9:57

gaming industry I think when I was like

9:59

in university like the gamer gate had happened

10:01

and just seeing not just

10:03

the culture of it, but also just the amount

10:05

of hours that people are

10:07

working there. I didn't

10:09

want to dedicate my life to games. I

10:11

didn't love it that much, unfortunately, and like

10:15

nothing against people who love it, like good for

10:17

you. But I knew that that wasn't my path.

10:19

So yeah, I just never ended up working in

10:21

the gaming industry. Welcome

10:30

to a world where your reading

10:32

list becomes your listening playlist. Introducing

10:34

listenlater.net, the new service that turns

10:36

articles into podcasts. Have you

10:39

ever stumbled on an interesting article but

10:41

just didn't have the time to read?

10:43

Maybe you're driving, working out or just

10:45

relaxing at home. With listenlater.net, email us

10:47

the URL and voila, our AI transforms

10:49

it into a captivating podcast narrated by

10:51

a voice that you'll love listening to.

10:53

And that's not all. listenlater takes your

10:55

inbox to the next level. Send us

10:57

any email and we'll convert it into

10:59

your personal podcast. It even handles

11:01

text from attachments, whether it's work

11:03

documents, newsletters, or that long article

11:05

you've been meaning to read, we've

11:08

got you covered. Crossing language barriers?

11:10

No problem. listenlater.net now offers multilingual

11:12

translation and narration. Choose your preferred

11:14

language and enjoy a world of

11:16

content right in your ears. Join

11:18

the revolution of convenience and versatility.

11:20

listenlater where your articles, emails, and

11:22

documents come to life and audio

11:24

form. Visit listenlater.net to start your

11:26

free trial today. That's listenlater.net to

11:28

start your free trial and transform

11:30

articles into your personal podcast. So

11:40

you worked at previous companies, cover, bench-side, Way Financial.

11:42

What's the journey that you've taken and how did

11:44

you end up at Netlify? Yeah, so

11:46

I kind of I did that in reverse orders. So started with

11:49

small business software because at the time when

11:51

I joined, that was kind of my quote

11:53

unquote area of expertise because I'd worked at

11:55

another competitor to Way Financial actually based in

11:57

Toronto called Freshbox early on in my career.

12:01

And the reason I kind of,

12:03

I jumped a little bit in different domains is

12:05

a little bit because I like

12:08

just exploring all these different problem domains. I

12:10

like just learning new things. When

12:12

I started my career, I wanted to be truly full

12:14

stack. I really liked the idea of jumping all over

12:16

the place. I love that there's so many different

12:18

problems in both the front end and the back end,

12:21

and I felt early on in my career that based

12:23

on what was needed in my various teams,

12:26

it tended, full stack tended to mean more

12:28

front end. And so my

12:30

first two roles, way financial and bench side kind

12:32

of gravitated toward that, even though there was a

12:34

little bit of backend development. But at the point

12:36

where I joined lever, I really wanted to do

12:38

more backend development. It was kind of a sticking

12:40

point for me that my next role at that

12:42

time was going to be back and focused. And

12:45

that's how I ended up on the API integrations

12:47

team at lever. And then

12:49

while I was there, I remember feeling

12:51

like in a way I was kind of still somewhat

12:54

doing the same problems over and over again, it was

12:56

always like the same, like,

12:58

and I imagine a lot of listeners who have a few years

13:00

under their belt working in software, you start noticing that

13:02

you're solving the same problem over and over again for

13:04

the same types of customers. So

13:07

I was telling a friend about this one day and

13:09

he was like, you know, you should try and do,

13:11

you should maybe look at like the developer tooling space

13:13

because it's very different. That's not

13:15

going to be the stuff you've been dealing with so far.

13:17

You're going to be working with a lot of people that

13:19

are like you and you'll end up learning a lot more

13:21

than maybe you feel like you're learning right now. That's

13:24

very different. So that's when I started looking

13:26

into that space. And fortunately for me, Netlify

13:28

was hiring at the time. And

13:30

so here I am today, almost two years later.

13:32

That's great. How has Netlify been for you

13:34

enjoying it? Oh, it's been really great.

13:36

It's been, and I'm not saying this because

13:38

I currently work there. It has been one of the best places

13:40

I've worked at. I find that

13:44

the people that I'm working with are, there's

13:46

so many experts in different domains that

13:50

I, especially the first like eight months, I

13:52

felt like such a dummy where I was like, yeah,

13:54

I'm coming in as a senior engineer and then there's

13:56

these people who have been like, you know,

13:58

reverse engineering for the first time. frameworks for how long,

14:01

like Matt Kane and some people

14:03

who have been working, like they just know the ins

14:05

and outs of how like CDNs work and all this

14:07

other low level stuff that I just took for granted

14:09

about building on the web. So

14:11

it was very humbling when I joined because

14:14

I was learning all these things that I

14:16

just did not think about that are like

14:18

fundamental building blocks of web development. And

14:22

I don't know of many other places that

14:24

would be as supportive as Netlify has been

14:26

with allowing me to pursue things

14:28

that I've always wanted to pursue, but it took

14:30

a bit of time like conference speaking,

14:32

which is how you and I met Josh. So

14:34

being able to do that last year was an

14:36

incredible opportunity. I'm forever grateful that Netlify let me

14:38

do that. And then just the problems that I'm

14:40

solving there, I've been able to work in so

14:42

many different areas that it feels like it's a

14:44

constant challenge every day. And I haven't been bored

14:47

since I joined. Well, there you go.

14:49

Your problem was solved. Yes. Is

14:59

your code getting dragged down by joins and

15:01

long query times? The problem might be

15:03

your database. Try simplifying the complex with graphs.

15:05

A graph database lets you model data the

15:08

way it looks in the real world instead

15:10

of forcing it into rows and columns. Stop

15:13

asking relational databases to do more than they

15:15

were made for. Graphs work

15:17

well for use cases with loss of

15:19

data connections like supply chain, fraud detection,

15:21

real time analytics, and gen AI. With

15:24

Neo4j, you can code in your favorite

15:26

programming language and against any driver. Plus

15:29

it's easy to integrate into

15:31

your tech stack. People are

15:33

solving some of the world's

15:35

biggest problems with graphs. Now

15:37

it's your turn. Visit neo4j.com/developer

15:40

to get started. That's neo4j.com/developer.

15:49

I want to jump into one of those areas

15:51

a bit because you've given talks, quite a few

15:53

talks in fact, about the edge. And

15:56

for some, the edge is still kind of a

15:58

nebulous term. Could you define for us? what

16:00

the edge is when it comes to computing. So

16:02

I'm happy to give my definition of it and it's the

16:04

best one I was able to come up with but I

16:07

also give the one that I've seen floating around the internet.

16:09

So the one floating around on the internet that I've seen

16:11

before, it's been around since about the 80s and

16:13

the edge is sometimes referred

16:16

to as the network edge. So for those who work

16:18

in an enterprise space, the network

16:20

edge is probably a term that you're

16:22

familiar with and that's where enterprise networks connect to

16:24

third party services. So it's

16:26

still a valid use of the term today

16:29

but when we're talking about the edge in

16:31

its modern context and we're seeing like Netlify

16:33

edge functions and Vercel edge functions when CloudFiller

16:35

workers come into the conversation, they're

16:38

referring to code and

16:41

locations that are executed much

16:43

closer to users than they maybe have historically

16:45

been. I like to joke

16:47

that on-prem is becoming a little bit

16:49

fashionable again in the form of edge. That's not

16:51

entirely accurate but to some extent it's a little

16:53

bit true because the edge

16:55

is entirely context dependent. The edge

16:58

depending on your these case could be the

17:00

phone that you hold in your hand for a

17:02

mobile application where the application is

17:04

maybe running some code or storing some

17:06

data on your phone as opposed to just storing everything

17:08

in a server far away or

17:11

it could be a webpage that

17:13

is caching stuff at

17:15

servers that are located by cell towers instead

17:17

of at a very distant server farm in

17:20

North Virginia or in Montreal, Canada. So

17:22

it is a very nebulous term and

17:25

the best way I can describe

17:27

it is whatever is closest to

17:29

your user depending on the

17:31

problem that you're solving or the use case that

17:33

you're dealing with. So

17:36

we have the umbrella term for edge which is as

17:38

you're describing, putting the logic, putting the

17:40

code in locations much closer to users

17:43

and you've listed a few of those

17:45

locations. The phone, the network edge closer

17:47

to say the cell tower, perhaps the

17:49

data compute if it's actually necessary to

17:51

be near there. If I were

17:54

a large enterprise with 9,001 different

17:56

types of

17:58

data and compute any data. happen? How do I

18:01

know which place to put which part of my

18:03

stack or part of my logic? Oh,

18:05

that's a great question. And I hate to

18:07

be that senior software engineer. That's like, it

18:09

really depends. So I've, in

18:11

my research for this, I've read

18:14

about larger enterprises like FedEx and

18:16

Walmart, the edge being their own

18:18

warehouses, and running operations from there.

18:21

And I'm not exactly sure what those

18:23

operations are necessarily. But that

18:25

is like one example of it. If

18:28

you're looking at something like, let's

18:30

say you're just, you have a

18:32

web application, you're modifying request to

18:35

response headers, or applying, let's say,

18:38

geo fencing, like you don't want to allow some

18:40

people in certain areas of the world, access

18:43

some content because of licensing agreements,

18:46

then you might be looking at data centers

18:48

that are run by your cloud provider of

18:50

choice, whether that's AWS, Google Cloud Platform or

18:52

Microsoft Azure. So yeah,

18:55

if you've got like 9001 data points, it's

18:57

hard to say necessarily what that looks like for

18:59

you, because it depends on the industry that you're

19:01

working in. Heck, I even remember speaking

19:04

with someone from a telecom company in

19:06

Australia that I'm talking to farmers and

19:08

like, there's like edge related to that. And I

19:10

wish I got a chance to talk to that person before

19:12

I hopped on this call, because I'm very fascinated that

19:15

you can imagine that the edge for

19:17

them might be the various

19:20

vehicles, it could be the warehouses that

19:23

they're storing stuff in, it

19:25

could be a lot of things. I

19:27

think one of the really common use cases that

19:29

gets brought up is edge functions. Can

19:31

you describe or define for us what

19:34

an edge function is in terms of,

19:36

say Netlify or typical web apps and

19:38

websites? Yeah, sure. So

19:40

edge functions are just imagine your

19:42

standard, I guess serverless

19:44

function, but rather than having it run at

19:46

a data center in an availability

19:49

zone, which is what you imagine being

19:51

the stereotypical data server farms. These

19:53

are functions that are run at servers

19:56

that are optimized to run

19:58

your code very, very quickly. they

20:01

tend to have much more

20:04

constrained hardware capabilities than the

20:06

standard serverless function, but they are run

20:08

physically much closer to your users, then

20:11

you imagine running that same function

20:13

in a in a distant server far away.

20:16

So from an end user perspective, what

20:19

is actually different or behaviorally different about

20:21

these edge servers compared to say traditional

20:23

Azure or AWS and Google Cloud hosted

20:26

cloud solutions? Well as far as

20:28

developers that are building stuff on

20:30

the edge, a lot feels

20:32

the same. There's some constraints

20:35

that are applied that involve let's say

20:37

like more limited memory, lower CPU time,

20:39

which is and I always need to

20:41

refresh my memory when I look up

20:43

CPU time versus wall time, CPU time

20:45

referring to the number of operations

20:47

you can do as opposed to like literally seconds

20:49

taking away on a clock. So there

20:51

are some some hardware constraints,

20:54

but generally speaking unless you're doing

20:56

something particularly intensive, moving

20:58

functionality to the edge generally feels the same

21:00

as if you were just writing normal functionality

21:03

in a data center, whether that's like normal

21:05

serverless functions or an edge function. Walking

21:07

through an example use case then, suppose that

21:10

I want to change some

21:12

text on a page for A-B experiments and I want

21:14

it to be very very fast and in

21:16

theory stretch goal it would be lovely

21:18

if I could base that A-B split

21:20

for people if they are logged in

21:22

not anonymous on their user data in

21:24

my old user data service. How

21:27

would that start to look like in

21:29

these new edge environments? That's

21:31

a great question. So assuming that the

21:33

user is already logged in and you're rendering

21:35

I guess custom content to this person, what

21:38

that might look like is and the

21:40

example I like to use is let's

21:43

say transitioning from a server-side rendered

21:45

page because I work in the web space

21:48

so that's what I'm familiar with.

21:50

So you have a server-side rendered page that

21:52

previously would maybe take the data that you

21:54

needed conditionally render the content and send it

21:56

back to the user and maybe you cache

21:58

that on the user's browser. When

22:00

you use the edge, what you can do

22:02

instead is you could take the common elements

22:04

of that page and make it static. So

22:07

now you can cache that at the CDN.

22:09

So in the way the CDN is

22:11

the edge, but it's just focused on storage,

22:13

whereas the edge is storage and compute capabilities.

22:17

So you've got the static value. And then

22:19

at the edge, what you can do is

22:21

basically whatever conditions you need to apply. So

22:23

using that user as an example, you know,

22:25

you want to show a certain

22:28

piece of content or some promotional item or

22:31

whatever your use case is, what you can

22:33

do is you can take the response, the

22:35

static response, and rewrite the HTML,

22:38

maybe update props if you're

22:40

using a React framework or whatever the

22:42

equivalent is in your framework of

22:44

choice. And then you can return

22:46

that response to the user. And what's also great, and

22:48

as I mentioned before, the edge also allows you to

22:51

cache stuff at the edge as well. So

22:53

you can cache that response going forward as

22:56

well for even faster response times going forward.

22:59

So if I were to try

23:01

to make that caching key dynamic,

23:04

could I then make it

23:06

completely cached for all anonymous users for speed's

23:08

sake? You could, yeah. And like,

23:10

if we're talking something generalized like a homepage,

23:12

that's very straightforward to do. If

23:15

you're doing something that

23:18

is highly personalizable, let's say you can

23:20

cache it, but your cache might not

23:22

be as long lived as you would

23:24

maybe ideally want it to be, just

23:27

because the cache size available

23:29

at the edge is much smaller than maybe what you

23:31

might have available in an AZ. That

23:34

being said, there's ways around

23:36

that. The talk that you mentioned before,

23:38

I referenced that there are regional edge

23:40

caches available, which they give you

23:43

a little bit of wiggle room so that you can

23:45

refer to these edge caches before you have to go

23:47

all the way to a distant AZ. So

23:50

there's that available to you. But generally

23:52

speaking, like, yeah, you can cache the personalized

23:54

stuff at the edge, but just be warned it

23:56

might not be as long lived as you would

23:58

hope it would be. Sure.

24:02

That's cool. That makes sense. I want

24:04

to switch topics a little bit because

24:06

you mentioned working on a feature a

24:08

bit back and that feature is of

24:10

interest. Private integrations. What are private integrations

24:12

when it comes to Netlify? Yeah,

24:14

sure. Netlify has always

24:17

had, or they've had for a very long time, this

24:19

concept of integrations. It's mainly

24:21

focused on hooking into

24:23

the build system. What

24:26

private integrations is, we're

24:29

trying to allow folks to do

24:32

much more on the Netlify platform than

24:34

just hooking into build events. We're trying

24:36

to make it easier for external developers

24:38

to maybe create custom tooling that leverages

24:40

more of Netlify's products, including Netlify functions,

24:43

the build events or the build hooks that I

24:45

mentioned before, and maybe going forward

24:47

other stuff as well. Private

24:50

integrations allows folks to package whatever custom

24:53

repeatable things that they need for their

24:55

sites, make an integration that

24:57

is available only to their team, and

24:59

be able to distribute it to all users

25:01

that use their Netlify team. We

25:04

think it's going to be pretty exciting for folks,

25:06

but maybe just have stuff that they're sick and

25:08

tired of everyone having to just run MPMI for

25:10

this one specific package. Updates are

25:12

just rolled out a lot more easily with the

25:15

use of something like private integrations. Do

25:17

you have an example of an update or a

25:19

set of things that someone would want to package

25:21

into an integration? Yeah, so the

25:23

first integrations that we've said, this is a public

25:26

integration, not private, but one of the first things

25:28

that we switched over was Sentry.

25:31

We have two different Sentry integrations and we

25:33

package them into one. But

25:35

something that we're, let's

25:37

say someone wanted to extend that further. Let's

25:39

say there's another feature of Sentry that's not

25:42

available in there, they don't want to wait,

25:44

and they also have some extra stuff that

25:46

they're doing on the side, maybe because they're

25:48

doing a migration between Bugsnag and Sentry. Someone

25:52

could package a private integration that

25:55

maybe sends events that they need to send to both

25:57

parties while they're in the middle of that transition. and

26:00

then allow people to install it on their sites as

26:02

they need it. I can see

26:04

this being the big time saver, as you

26:06

mentioned, especially for larger teams, where their entire

26:08

teams are sections of dedicated to helping other

26:11

teams spin up new products or services. Yeah,

26:13

yeah. That's the hope, anyway, is just kind of

26:16

remove a lot of the toil that we

26:18

sometimes deal with. Let's hold in on

26:20

that in the general sense. The industry

26:22

we're in programming always has these cycles

26:24

come through that, in retrospect, seem obvious,

26:26

but in foresight are hard to detect.

26:28

We keep making things that were difficult

26:30

easier. We're automating, we're finding better ways

26:32

to do things, and we've already touched

26:34

on two points for that. The edge,

26:36

which makes it easier to put stuff

26:38

closer to users and have more finely

26:40

tuned computing, and then things like integrations,

26:42

which quite a few companies are doing

26:44

something like the ability to package little

26:46

reusable chunks of logic or code. Do

26:49

you have any other particular areas in Netlify you're

26:51

excited about on that trend of making things

26:53

easier for people? Oh, personally,

26:56

and I know this isn't like

26:58

a Netlify roadmap item, but for

27:01

me personally, I want to make

27:03

it really easy for folks to

27:05

create more environmentally conventional software is the best

27:08

way I can describe it. I

27:10

saw Sarah Bergman do a talk at QCon London,

27:12

I think it was last year, and she talked

27:14

about the Green Software Foundation and what they're pushing,

27:16

and there were other talks like it, but I

27:20

got really motivated when I realized

27:22

that I work in a space

27:24

where developers are interacting with us

27:26

all the time, and

27:28

we're helping them build software

27:30

ideally faster, easier, more

27:33

bug free, and at least

27:35

for me, I'm now having, I've

27:38

seen the light a little bit where it's like, yeah, this is

27:40

a huge problem. I'm trying my best to keep

27:42

my eye out for opportunities where

27:45

I can highlight some of the things people can do to

27:49

improve the, I

27:51

guess, almost the energy efficiency of their software

27:54

or at least highlight things for

27:56

people to be mindful of as they're building

27:58

software so that we can try the

28:00

carbon footprint of the

28:02

things that we build and that people are using every day. Do

28:05

you have an example of a good technique

28:07

that can help folks build more efficient green

28:09

software? Well, I'm going to sell

28:11

the edge on this right away. The

28:14

edge is one area where people can

28:16

look to adopt it for maybe some

28:18

high traffic, but very focused pieces of

28:20

functionality. So I mentioned before about setting

28:22

requests and response headers, maybe split testing

28:24

is another great example, setting session cookies.

28:26

They're all things that we

28:28

do on thousands, hundreds of

28:30

thousands, sometimes millions of

28:32

requests. And even just

28:34

like small little things that move to the

28:37

edge. And if you're caching very popular requests

28:39

at the edge, the reduced energy

28:41

that's consumed from transmitting that data can

28:43

add up to something really meaningful. And

28:46

I know generally measuring these in

28:49

more concrete terms can be a bit

28:51

challenging because you've got geographic considerations, you've

28:53

got machines that people are using to make

28:55

the requests. You've got cloud providers

28:57

that have various pieces of hardware running that all

29:00

tie into this. And so I know the hard data

29:02

is sometimes a little bit hard to come by. But

29:04

generally speaking, it's safe to say that the

29:07

most environmentally friendly thing is to not run

29:09

things if you can at all avoid it,

29:12

or reduce the amount of machines that

29:15

you're running. And if you have to run these things,

29:17

make sure that they're fulfilling it as

29:20

fast as possible or reduce the amount

29:22

of machines needed to fulfill that overall request.

29:24

And so that's where the edge can be

29:26

kind of helpful as it can reduce that

29:28

need maybe for bigger databases potentially on your

29:30

use case, reducing the distance that

29:32

the request has to travel, which means not just better

29:35

performance for your users, but a reduced

29:37

environmental footprint because the resources that are

29:39

making requests aren't waiting idly for as

29:41

long and consuming electricity, just waiting for

29:43

a response to come back. So

29:45

it adds up to little things. But I think that's one thing

29:47

that people can do to at least

29:49

start moving in that direction. And hey, you get

29:52

better performance out of that, right? Which is win-wins

29:54

all around. That's a win for

29:56

the users win for whoever has to foot the bill

29:58

for your servers. Exactly. It's

30:01

funny, there's been a lot of talk about

30:03

accessibility, about how there is a concept of

30:05

this electronic curb cut, things that were maybe

30:07

originally intended for one segment of users, for

30:10

physical curbs that was folks in wheelchairs, and

30:12

are actually also very beneficial for other segments,

30:14

such as people with a lot of shopping

30:17

carts or wheel strollers. And

30:19

then what you've just brought up strikes me as a

30:21

parallel to that, where one avenue of

30:23

exploration could be trying to make more green

30:25

software for the very noble good cause of

30:28

destroying the planet slightly less. But also

30:30

there is the really strong business use case

30:32

of paying less for servers and giving it

30:34

better experience for users. It's kind of fortuitous

30:36

that those all kind of blend together and

30:39

work well. Yeah, and I think it's

30:41

worth calling out too, like just from

30:43

an accessibility standpoint, I think one

30:45

thing that we do need to

30:47

note is depending on, especially for bigger companies that

30:49

are operating in different places all over the world,

30:52

I think one thing that especially folks

30:54

like me living in North America with

30:56

pretty reliable internet access, that's not always

30:59

true for a lot of people. And

31:01

something like the edge where you focus on

31:03

running things as close to the user as

31:06

possible in a way that they don't need

31:08

to rely on constant internet access for the

31:10

functionality that they need to work

31:13

reliably. I think that

31:15

that's something that is worth keeping in

31:17

mind, because yeah, depending on who your

31:19

users are, your business product may fail

31:22

if your users are folks

31:24

that live in areas with unreliable internet.

31:27

And so I think the edge can help at

31:29

least make some software much more accessible than it

31:31

has been previously. Let's talk

31:33

about that a little bit. Suppose I'm

31:35

in rural farmland or a city with

31:38

overworked overloaded ISPs. What

31:40

about the edge makes it so that I would

31:42

be able to better use a company's services and

31:44

ideally also pay for those services as a user?

31:47

Yeah, so there are some design patterns

31:49

out there that leverage this a little bit.

31:51

And there's also some products out there that

31:53

they are a form of cloud computing on

31:56

the edge, but they're targeted toward people that

31:58

are entirely remote. talking about

32:00

the design pattern first, there's,

32:03

I saw this in a talk, his name

32:05

was Carl's Fair, I think, I hope I pronounced his name

32:07

right. If you're, if you're listening, Carl, that

32:09

talked about offline first architectures. So

32:11

the idea being that you have

32:13

this device or you have the

32:15

server that all these other,

32:17

I guess, nodes, so other applications, other

32:20

machines are making a request to that

32:22

are updating data. And the

32:24

requests aren't immediately going to the

32:26

remote or to the AZ data

32:28

center that's storing the majority of

32:30

data. Instead, it happens every

32:32

so often, or when the network

32:35

becomes available again. So that

32:37

is one approach where the edge can

32:40

help keep things accessible. And let's say

32:42

you're overloaded ISP example, where you just

32:44

wait until something becomes available, you try

32:46

a few times, and then eventually it gets going. But

32:48

at least the people that need a

32:50

response now are able to continue what

32:53

they need to be doing. The

32:55

other bit with hardware, and this

32:57

like blew my mind when I found out about it, I

32:59

didn't even know about it until someone that I used to

33:01

work with who worked at AWS briefly told me about it.

33:04

There's devices out there that

33:06

in this case, AWS snowball that

33:09

it gets shipped to you and you just like hook

33:11

it up to your servers and it gives you cloud

33:13

computes, like right on site. And

33:15

so you can keep it around to run

33:18

cloud compute functionality, but you can also use

33:20

it in situations where you maybe need to

33:22

migrate a lot of data and you're a

33:24

bit constrained by the bandwidth and being able

33:26

to upload the data fast enough, given your

33:29

intermittent or non-existent internet access. And

33:31

so in such a case, you would just load it

33:33

onto the device, ship it back and it gets uploaded

33:35

to the cloud for you. And there's much more like

33:38

scaled up versions of that storage thing. I think

33:40

the one that most people know about is like

33:42

snowmobile or something where it's the giant transport truck

33:44

that gets driven out. But snowball

33:46

gives you compute capabilities, not just storage. So

33:49

it's UPS for your data. Basically, I know

33:51

there's some joke on the internet around like

33:54

NASA and cassettes in a station wagon or

33:56

something that it's kind of that problem. That's

33:59

the funniest thing. We've gotten so advanced at

34:01

storing our data that we can now ship

34:03

exabytes of it manually

34:05

to your needs. Exactly. I

34:09

love it. So, green software, a highly

34:11

efficient software, sustainable software, this is a relatively newer

34:13

area and at least public discourse, although I know

34:15

quite a few folks have been working on this

34:17

for a while. Is there some

34:19

kind of sustainability guideline from a larger consortium that

34:22

we can refer to for how to build this

34:24

kind of software? Yeah, I

34:26

actually found out, I think it was

34:28

in last September. There was

34:30

the W3C published these web

34:32

sustainability guidelines that people can refer to

34:35

and they cover sustainability from

34:37

a number of different angles. Accessibility

34:39

is one of those components, environmental is another,

34:42

social equity is another and that doesn't cover

34:44

the whole list obviously, but that's one

34:46

document that's pretty easy to digest. It's

34:49

not, when I read it, it wasn't overly dry.

34:52

Like I know some specifications can go on for a bit,

34:54

but these were very short and to the point and very

34:56

helpful. And although it's not a

34:58

spec, I would encourage people to maybe take

35:00

a look at the Green Software Foundation that

35:02

I mentioned earlier because they do have design

35:05

patterns and resources that people can refer

35:07

to that also can help guide people

35:09

toward making more efficient software

35:11

that also will end up producing your

35:13

spend more than likely. That's

35:16

great. I want

35:18

to end the interview with a few fun questions.

35:20

We started talking about your dog. Could you just

35:22

give us a little more information on your dog?

35:25

Oh yeah, I would love to. So

35:28

her name is Ada. She is not

35:30

quite two years old yet and she's

35:32

one of those poodle mixes. I know

35:34

they're quite popular. She's

35:36

like Bernice Mounted Dog Poodle Makes it,

35:38

honestly. She looks all poodle. She's got

35:40

legs for days. And

35:42

yeah, she is a joy. She likes to sit

35:44

at my desk when I'm programming and somehow

35:46

get involved in all my meetings. And

35:49

yeah, she's just in general a very happy bean. If

35:51

folks want me to post more pictures on Twitter, you're going

35:53

to have to ping me for them. But

35:56

she also has her own Instagram if people really if

35:58

they want to follow her. It's like. data.beans with

36:00

a Z. Completely or

36:03

mostly unrelated, you have something

36:05

called the 418. What is the

36:07

418? Ooh, I'm so

36:09

happy you asked me about this. So the 418 and

36:11

for the folks who are very familiar with

36:13

the HTTP status codes is like that teapots

36:15

one, which I thought was a nice fun one to

36:18

name Azine, which is what this is after. The

36:20

418 came out of this idea that I've been

36:22

kicking around in the back of my head early

36:24

last year that I told a

36:27

friend of mine, Christina Berger, who is my co-founder in

36:29

this, about this idea where

36:32

I wanted to capture that

36:35

excited, inspired feeling that I was

36:37

getting from interacting with so many

36:39

awesome people at conferences, where

36:41

we're not just trading ideas about,

36:45

you know, what cool things technology is doing

36:47

in our lives, but also just

36:49

the fun hobbies that we were having and how

36:51

they may be, whether or not they tied into

36:53

technology, but just generally just having fellow

36:55

tech people that we just were

36:58

excited about different things in life. And

37:01

in general, like it tied into this feeling

37:03

to where the traditional tech meetups didn't really

37:05

feel like my place. Like I just never

37:07

felt like I fit in. And

37:11

I pitched this idea to Christina, like,

37:13

Hey, what if we take a page

37:15

out of this playbook that goes

37:17

back to the, at least mine exposure that

37:19

was the nineties with the riot girl movement,

37:21

I had read a book a few years

37:23

ago about how zines were part of that

37:25

culture with trading ideas

37:28

and art and building a

37:30

community across geographic distances. And

37:33

why don't we do a essentially attack version

37:35

of that. And the four

37:37

18 was born. So we're actually

37:39

like really, really close to putting out

37:41

our first issue. We got a couple of

37:43

features from some fantastic people

37:45

in the industry. And

37:48

we also added a few fun things in there. I

37:50

even built a crossword puzzle for this, which I was

37:52

like, so excited to do. And I hope that the

37:54

people who do do it have a good time with

37:56

that. So I think we need to clarify

37:58

what is a zine and is it. not pronounced

38:00

design. I'm gonna say zine. I'm not

38:02

going to get into a holy war of whether it's a zine

38:05

or zine. The zine is a much

38:07

more so you imagine a magazine is super glossy.

38:10

It's professional. It's got like, you know, a team

38:12

of 20 people or so working on it. Zine

38:14

is a much more small scale,

38:16

homespun version of that. It's often

38:19

people who are artists that

38:21

are maybe trying to you know, show a

38:23

story or something and they're drawing it out

38:25

themselves or you know, way back in the

38:27

day, some people would like cut out magazine

38:29

things, glue stick them on paper and then Xerox

38:31

and send it off to people. And

38:34

they also have some roots in

38:36

exchanging, you know, really ambitious

38:38

ideas, stuff that maybe wouldn't get published in

38:40

mainstream publications. And they have a really fascinating

38:42

history to them where I think they date

38:44

back like as early as the 1940s. And

38:48

they provide a picture for people that do study

38:50

history of what life is like in different communities,

38:52

because this was the way that they communicated was

38:54

maybe to the circulation of zines.

38:57

And so, yeah, they're everything

38:59

from like, you know, art

39:02

projects to political statements to

39:04

just sharing a love of a particular niche

39:07

interest like gardening, it can be so many

39:09

things. So honestly, if people

39:11

wanted like just a sense of just how many

39:13

there are in the world, go to like XZ

39:15

zine category, because there's so many different things out

39:18

there. And that's like just touching the surface. So

39:20

the 418 is a zine about tech

39:23

and adjacent areas? Yeah,

39:26

so I guess our slogan on is like

39:28

thoughts on tea and tech, although we

39:30

don't focus too much on tea. But our general

39:33

theme is what are the things that we would

39:35

talk about with our friends over tea? What

39:37

would make us excited? What would we have

39:39

fun doing with our friends? And that

39:42

could be, you know, talking

39:44

about this new recipe we're trying

39:46

or working on a puzzle together, which I

39:48

think that might be a very Canadian thing.

39:50

I don't know. There's like this trope of

39:52

people who I don't have a cottage, but

39:54

for people who do have cottages, they go

39:56

up in the summer and they have this

39:58

big newspaper crossroads that everyone works on over

40:00

the course of the weekend. Yeah,

40:03

so I don't want to spoil too much of what we've

40:05

got in there, but it's meant to be

40:07

things that you would do with friends and

40:09

exchanging ideas that we get really excited about

40:11

with tech, that like, that keeps us excited

40:13

about all the things that are happening in

40:15

the industry. Because it's not necessarily

40:17

the newest framework, but it could be how

40:19

tech is being applied in a particular area

40:22

of our lives. Well, that

40:25

might have invalidated my next and last area of

40:27

question, but I want to talk to you a little bit about that,

40:29

what's coming next in tech area. You've looked

40:32

at and worked with a few different areas,

40:34

more recently, Netlify, Edge, Private Integrations, and we've

40:36

talked a bit about how things are making

40:38

it easier for folks to develop. What excites

40:40

you about the next five to 10 years

40:43

in tech? And where do you see your

40:45

personal areas going with that? Oh,

40:47

that's a hard question. I sometimes don't even

40:49

know what I'm doing next year, sometimes in

40:51

tech, because it changes so quickly. I

40:54

think at least, I can't speak

40:56

to 10 years, but at least for the next five

40:59

years, I am

41:01

really excited to see the focus

41:05

on the impact of software that we're starting to see.

41:08

And there's just like with

41:10

AI and things being able to run much

41:13

closer to people, a much more

41:15

powerful hardware. I imagine, I

41:17

can't even imagine the different

41:20

things that are gonna come out as a result

41:22

of all that power basically being available to us.

41:26

At least for me, I'd love to continue

41:28

working with developers and

41:31

chase my newfound passion to like figuring out

41:33

how to maybe push green

41:35

tech forward, because I think

41:37

it's super important, especially right now. And

41:40

yeah, I hope I get the chance to make an impact

41:42

there over the coming years. Well, that's

41:44

an excellent note to close out on. Erica, this

41:46

has been fun. Is there anything else you would

41:48

want to cover or mention before we hang up?

41:51

No, I think that's it. Thank you so much again

41:53

for having me on the show. It's been a

41:55

pleasure chatting with you. Cool. Well,

41:57

with that, this is Josh with Software Engineering Daily. Thanks.

42:00

and benefit