Liz Fong-Jones (方禮真) Profile picture
Jul 26, 2018 83 tweets 32 min read Read on X
Livetweeting of the #GoogleNext18 developer keynote will commence here in this thread around 9am PT. So excited for @mmeckf and everyone else presenting! You can also watch live at g.co/nextonair
And here comes @adamse! He's polling the audience to find out how many people are developers? operators? gophers? #GoogleNext18
"Setting up a server is an API call now." "We don't customize and configure individual servers any more, we use containers."

"The role of developers is changing." #GoogleNext18
Rolling a video now of some amazing developers of color in Nigeria, India, Canada, Brazil, and more! #GoogleNext18
.@adamse's 3 promises: (1) deliver innovation, (2) commitment to open source and open cloud, (3) be a part of the developer community #GoogleNext18
Next up: @mmeckf to talk about tooling. She's passionate about developers because she is one. And she used to work in ops and carried a pager. "It works well on my machine." #GoogleNext18
She leads developer and operator tools at @GCPcloud. DevOps = (1) end-to-end tooling from source to test to observability, with culture.

Automate away the toil and making the humans happy. #GoogleNext18
Humans develop and operate our services. But we have way too many choices. to think about. Google's goal is to make the DevOps tooling decisions easier, so people can get to bringing their ideas to life quickly. #GoogleNext18
DevOps = faster deployment, with security. 46x faster releases, 7x less likely to break services with DevOps culture. (citing @nicolefv research)#GoogleNext18
Three tips: (1) use data to guide you, (2) have shared fate, (3) have transparency rather than secrecy and siloing. Integrate o11y into apps. (4) embrace blameless postmortems and continuous improvement. #GoogleNext18
Easy to say, but living it is magic. It builds trust between developers and operators.

Today's agenda: code, build, deploy, and monitor. #GoogleNext18
Cloud Source Repository, CloudBuild, and Stackdriver workflows, demoed end to end.

Find a problem with Stackdriver, search for the code, commit a fix, and commit. Live demo by Morgan Mclean. #GoogleNext18
Stackdriver Trace to find the slow spans -> Stackdriver Profiling with flame/"icicle" graphs. #GoogleNext18
"Look at the blame layer." Oh hey, Morgan introduced the code to slow it down for the demo. Fortunately there's a pull request that fixes it, and cloud build run for the pull request. #GoogleNext18
We need to make sure that our pull requests aren't broken. Cloud Build and Github are partnering to put CI right within the GitHub UI. Any Dockerfile in the root of your Git repo will allow one-click verification of your build. #GoogleNext18
You can see the logs of the cloud build live as it builds and pushes to Google Container Registry. #GoogleNext18
We also have security scanning built in so you can't accidentally push a vulnerable build.

And then Spinnaker will deploy your image for you. #GoogleNext18
Spinnaker does automatic canarying and canary analysis before promoting the canary to all remaining instances. Making it easy and automated eliminates human bias and ad-hoc steps, minimizing deployment risks. #GoogleNext18
More on open cloud: collaboration with Terraform, Gitlab, and more. #GoogleNext18
Trace, Profiler (saving 10% of compute bill with an hour of work), Cloud Build (and vuln scanner), Spinnaker for deployments & canary analysis. #GoogleNext18
Google's wisdom of 2 decades of developer and operator tools now are available so that you can scale and speed up your development. #GoogleNext18
Next up: @kelseyhightower. "I've wound up using k8s so long I finally ended up in a container." #GoogleNext18
How can we put together all of our technologies, starting from 0 with a database to an API endpoint, in 15 minutes.

"Everyone like terminals, that cool?" #GoogleNext18
We'd like to take the locations of Google datacenters in the US and get the weather at each of them. But we need coordinates for the text strings to get weather.gov to talk to us. #GoogleNext18
"If you haven't hear of Stack Overflow you got no clue what's goin' on. High quality code there. I'm not sure who this @tpryan guy is..." --@kelseyhightower #GoogleNext18
So let's just grab this code, but we want to make sure it doesn't contain any security vulnerabilities... "Good programmers copy, but great programmers paste!" --@kelseyhightower #GoogleNext18
So let's start with a CloudSQL postgres database, using the Google Cloud Console UI to create a database and user... #GoogleNext18
Initialize the development environment, set up our database, and then build our application that we got off Stack Overflow, which compiles and got us the weather data into our database. Yay! #GoogleNext18
But we might want to run it in a cluster, rather than running it locally. "There's no better place to run your code than a k8s cluster. There, Marketing, I did my job." --@kelseyhightower #GoogleNext18
So we have two different node pools - one with 3 nodes running envoy that's acting as our frontend reverse proxy pool that GCLB points to, and the rest that are our web tier. #GoogleNext18
"The price of admission to k8s is a container image." We need to build the binary, then have cloud build make the docker image for us.

Our cron job will automatically deploy using kubectl. #GoogleNext18
Because of VPC, if you're developing within GCP, you can just directly connect to k8s pods using the same IP address k8s shows you. #GoogleNext18
So our app is now up and running! But what if we need to debug it? OpenCensus lets someone else who isn't intimately familiar with the app architecture to examine the state of the app e.g. by sending traces to Stackdriver Trace and visualizing them there. #GoogleNext18
Annotations let us see the *exact* SQL query being run, rather than just the start/end times of the span. #GoogleNext18
And just to add some shiny: let's add some DialogFlow to our application so we can ask the weather assistant out loud for the weather. #GoogleNext18
and /healthz makes an appearance as we verify that envoy is working and our dialogflow deployment is working correctly. #GoogleNext18
Give it a few examples to learn from and some synonyms, and point it at our API backend we developed earlier. #GoogleNext18
"Looks like you have a pretty large cloud here, I hope the demo gods are on your side." says the Assistant! and it works, and there are live logs to prove it!. #GoogleNext18
Next, here's @the_thagomizer to talk about languages and giving back to our tech communities! #GoogleNext18
"I'm a giant language nerd," says @the_thagomizer. So let's talk about @golang!

There are 1M gophers worldwide! and it's the most talked about language in DevOps. #GoogleNext18
It's the language of the containers world. Everything Kelsey showed was written in Go, including k8s and docker. And it has first-class concurrency. #GoogleNext18
introducing the Go Cloud Project, allowing for cloud-agnostic APIs rather than having to translate each API's inputs and outputs for each cloud provider yourself. #GoogleNext18
Eno is demonstrating uploading to both GCS and to S3 from the same application code, changing from a big gross mess of conditionals. "We can do a lot better" by separating the application logic and our dependencies. #GoogleNext18
Next, Javascript. Serverside javascript using node.js is awesome! And there's more stuff like robotics you can use it for.

You can use your favorite npm modules on GAE with the nodejs runtime. #GoogleNext18
node.js runs on every service GCP offers.

We put our code where our mouth is by contributing to languages and surrounding ecosystems, and have a large Cloud Developer relations team (waving at those of us in red). #GoogleNext18
Next up, Sam and Francis from Firebase on mobile development! #GoogleNext18
Let's say we have a banking app and we want to add a mobile UI to it. Firestore lets us directly access the remote data using the Firestore API as if it were a local database. #GoogleNext18
and it has realtime update propagation, ensuring web and mobile apps show the same thing in realtime. #GoogleNext18
Firebase authentication rulesets ensure single signon and enforcement of ACLs, without having to write your own proxy or auth service. #GoogleNext18
We can write a billpay function on the server side and deploy it using Firebase Cloud Functions. The auth state is seamlessly passed along from the client. #GoogleNext18
On improving app stability and performance: 42% of 1-star reviews are because of bugginess and slowness; 73% of 5-star reviews are because of polish. #GoogleNext18
Firebase improves app quality by letting us find out about crashing in realtime rather than through negative Google Play Store reviews. #GoogleNext18
Slack integration with Crashlytics lets us find out in realtime about crashes [ed: although maybe file a ticket so you can track things]. #GoogleNext18
Firebase has integration with BigQuery so you can perform deeper analysis and integrate other data sources.

And you can also do A/B testing within Firebase without needing to deploy new app versions. #GoogleNext18
Firebase Predictions adds ML to allow mobile app developers to, for instance, target notifications to the right audience e.g. send bill pay reminders to people who are likely to be late otherwise. #GoogleNext18
Next up is @yufengg, but unfortunately due to an internet glitch I wasn't able to livetweet the first part of his talk. Many different APIs provided by Google for automated object/scene detection, speech recognition, and more. #GoogleNext18
"Be a good steward of your data, and it will reward you." --@yufengg #GoogleNext18
When we want to train data, we can use TensorFlow locally on a subset of our data, but we can bring our models to our large quantities data in the cloud using Cloud ML Engine, and try different approaches and different datasets. #GoogleNext18
You have access to CPUs, GPUs, and TPUs on GCP.

And Cloud ML Engine lets us deploy to production easily. #GoogleNext18
Next up: Anthony Goldbloom, CEO of Kaggle. Kaggle is the one destination for the ML community -- and not just competitions, but also learning resources, open source kernels, and datasets. #GoogleNext18
[ed: open source datasets, and kernels for editing above ^^ not very versed in this ML stuff]

Kaggle kernels provide an environment to allow cloud editing in python or R in multiple different IDEs, and allows 6 hours of CPU by default for free. #GoogleNext18
Collaboration on shared datasets is important to advancing ML research. #GoogleNext18
Having access to shared notebooks, including both notes and individual cells is great. [ed: this is reminding me of my Mathematica-using days in college the first and second times around...] #GoogleNext18
[ed: except for it's collaborative, cloud-enabled, and allows remote execution of all kinds of analytics workloads.] #GoogleNext18
Next up: @samok and @unity3d on how GCP helps startups. #GoogleNext18
[ed: as a former game developer, really excited about this. maybe I'll get to volunteer to help @puzzlepirates to get case studies for my SRE devrel work in the coming months...]

@suhaildutta is here to tell us about how unity enables democratizing development. #GoogleNext18
90% of the top mobile apps in Google Play are connected games, but delighting users is hard and requires a lot of infrastructure. #GoogleNext18
Unity is developing joint products for developers running on GCP, and also moving its own infra to GCP. #GoogleNext18
Back to to @samok: GCP provides discounts to startups in early phases for critical collaboration and development products, and also connects people to community of other startups. #GoogleNext18
You can get started at cloud.google.com/startups. Startups across all sectors are using GCP. First case study: @ceejbot of
@npmjs #GoogleNext18
NPM powers the largest registry of open source software. Any modern frontend framework uses `npm install`, and people don't even think about what happens when they just run it.

NPM used GKE to produce an easy to package and use enterprise version of its product. #GoogleNext18
"Just saying k8s" is not an answer. You have to install, manage, and operate it, and scale it out to demand. And your customers will need to do that if you're operating SaaS. #GoogleNext18
GKE had exactly the API npm needed. "Give me a cluster, here are my compute and memory requirements."

and GCS was right there for backing storage. #GoogleNext18
Customer resources stay isolated in separate tenant projects. 3 developers at NPM were able to develop NPM enterprise in months. #GoogleNext18
"You don't need to hire a staff that are k8s experts to use k8s with GKE." #GoogleNext18
Case study 2: Lina Nilsson of Recursion Pharmaceuticals talking about redefining drug discovery. Drug discovery costs are going up by an order of magnitude, not down, instead of following pattern of compute which goes down over time. #GoogleNext18
Serial formula: 1 disease, 1 hypothesis, a decade per study. Instead, new formula of testing many diseases and millions of hypotheses in parallel. [ed: curious how this avoids the problem of finding false p<0.001 findings when you don't have a hypothesis in advance] #GoogleNext18
Robots solve the problem, running 100k experiments a week and generating 2M images per week and 1PB of data per month. #GoogleNext18
Cloud ML allows for prioritizing which experiments to run that have the highest chance of success. #GoogleNext18
[ed: ah, now I see. they're designing the experiments in advance, so they know what they're testing for before running. still, I imagine they have to re-run experiments that turn up successful given the high numbers involved]. #GoogleNext18
Recursion already has 1 drug in trials within 5 years of existing and a robust pipeline in parallel behind it. #GoogleNext18
Wrapping up: @adamse with 500 local devfests, thanking the audience for being enthusiastic and engaged.

The next NEXT is April 9-11, 2019 in SF. #GoogleNext18
Next up: @sarahnovotny talking about having people vote using Dory with panel questions for herself, @bradfitz, @kelseyhightower, Eric Brewer, and Jia Lee. [ed: I gotta dash to @btreynor's session, but it was great to livetweet the keynote and hope you enjoyed it!] #GoogleNext18

• • •

Missing some Tweet in this thread? You can try to force a refresh
 

Keep Current with Liz Fong-Jones (方禮真)

Liz Fong-Jones (方禮真) Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

PDF

Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @lizthegrey

Oct 3, 2018
Final talk I'll be getting to at #VelocityConf before I dash to Toronto: @IanColdwater on improving container security on k8s.
@IanColdwater She focuses on hardening her employer's cloud container infrastructure, including doing work on k8s.

She also was an ethical hacker before she went into DevOps and DevSecOps. #VelocityConf
She travels around doing competitive hacking with CTFs. It's important to think like an attacker rather than assuming good intents and nice user personas that use our features in the way the devs intended things to be used. #VelocityConf
Read 36 tweets
Oct 3, 2018
My colleague @sethvargo on microservice security at #VelocityConf: traditionally we've thought of traditional security as all-or-nothing -- that you put the biggest possible padlock on your perimeter, and you have a secure zone and untrusted zone.
@sethvargo We know that monoliths don't actually work, so we're moving towards microservices. But how does this change your security model?

You might have a loadbalancer that has software-defined rules. And you have a variety of compartmentalized networks. #VelocityConf
You might also be communicating with managed services such as Cloud SQL that are outside of your security perimeter.

You no longer have one resource, firewall, loadbalancer, and security team. You have many. Including "Chris." #VelocityConf
Read 19 tweets
Oct 3, 2018
Leading off the k8s track today is @krisnova on migrating monoliths to k8s! #VelocityConf
@krisnova [ed: p.s. her ponies and rainbows dress is A+++]

She starts by providing a resources link: j.hept.io/velocity-nyc-2…

The problems we're solving:
(1) why are monoliths harder to migrate?
(2) Should you?
(3) How do I start?
(4) Best practices #VelocityConf
.@krisnova is a Gaypher (gay gopher), is a k8s maintainer, and is involved in two k8s SIGs (cluster lifecycle & aws, but she likes all the clouds. depending upon the day). And she did SRE before becoming a Dev Advocate! #VelocityConf
Read 29 tweets
Oct 3, 2018
Final keynote block: @lxt of Mozilla on practical ethics and user data. #VelocityConf
@lxt And also ethics of experimentation!

"just collect data and figure out later how you'll use it" doesn't work any more. #VelocityConf
We used to be optimistic before we ruined everything.

Mozilla also used to not collect data, and only had data on number of downloads, but its market share went down because they weren't measuring user satisfaction and actual usage. #VelocityConf
Read 25 tweets
Oct 3, 2018
Next up is @mrb_bk on why marketing matters. #VelocityConf
@mrb_bk Hypothesis: marketing >> code in terms of software adoption. [ed: and this is why I became a developer advocate!] #VelocityConf
You need to consider community early when developing a product.

Always ask, "Why do people matter?" "Why does adoption matter?" #VelocityConf
Read 17 tweets
Oct 3, 2018
Next up is @rogerm on O'Reilly's insights into trends with Radar. #VelocityConf
@rogerm They look at changes in search terms year on year; the two largest increases are k8s and blockchain. #VelocityConf
People are becoming less interested in broader topics and more interested in specific technologies e.g. pytorch. #VelocityConf
Read 5 tweets

Did Thread Reader help you today?

Support us! We are indie developers!


This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Don't want to be a Premium member but still want to support us?

Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal

Or Donate anonymously using crypto!

Ethereum

0xfe58350B80634f60Fa6Dc149a72b4DFbc17D341E copy

Bitcoin

3ATGMxNzCUFzxpMCHL5sWSt4DVtS8UqXpi copy

Thank you for your support!

Follow Us!

:(