Node Day 2015

##Experiences at Conde Nast

1. Architecture based on each teams - Epicurious, etc

2. Greenfield rebuild

3. Jenkens for CI

4. Monitoring uses New Relic (alternative data dog)

5. Deploy (AWS CodeDeploy for zero down time deploy) – capistrano (rails) and fab are alternatives

6. Microservices mappings entitlements versus authentication in nginx

7. Logging – bunyan for node and logs in json, use logstash and elastic search to combine logs from microservices

8. Systemd (systemd is native to centOS similar to redhat)

9. Puppet for configurations

10. Private npm registry through nodejitsu (FTW, early adoption before npm private registry+enterprise is available)

##Panel, stories from conde nast, Bloomberg Web, Dow Jones (parent of Wallstreet Journal which is running on node)

1. Schism between node.js and io.js

2. Date warehouse, data science and sentiment analysis in a lucene search engine?

3. Serverside vs clientside javascript – memory usage and debugging through new relics

4. EC6 is coming in June! Not yet adopted in Production, dev experiments in babeljs and typescript

5. Bloomberg’s deployment is also private and on private infrastructure, WSJ is on AWS

##IOT and node (spark.io and Microsoft)

• 8-bit+ chips are capable of running node

• Democratization of chips as they come down in cost

• Private cloud or bring your own cloud to improve security, private data

##NearForm experiences with Node since 2011

• Since Jan 2011 the number of node modules has grown exponetially to ~140000 modules

• Isomorphic javascript

• Private repositories where developers can publish modules

• Greenfield outside of startups are rare and enterprises involve legacy migration which is brittle over time

• Keep processes small, week worth of bad code, microservices of node services versus a single monolithic process

• Rapid, Less code – value sooner, 2x faster to deliver to java and .net

• 2 year +, once the problem space has been mapped out we can move to scala or go so it can scale better

##Enterprise Adoption Panel (ancestry, fidelity, and paypal)

• At paypal – large private npm reigsry (early adopter), krakenjs can handle large number of hits framework over express, expressive more lines of README than code

• Ancestry is a java and .net shop that has been converting over from monolithic microsoft model

• Adoption was over time, tested in the UI space first then organically spread but not without resistance. 2x rapid delivery convinced some.

• Start out small and scale out at ancestry, 5 people week/up to about 50 people out of 120

• Support experience at fidelity - first team on node sometimes can get stretched out thin responding to 5 other teams regarding node questions even for well documented libraries like express or yeoman

• Internal open source model at ancestry and fidelity – default public, writing code with the mindset to open souce, external open source model at paypal

• Git adaoption - atlassian stash at ancestry and fidelity, github enterprise at paypal (all three companies use git).

• Standardardization for documentation

• Default public model in repositories (compared to default private model previously which were siloed)

• Start up small and test out the problem space, go slow to figure it out and then iterate on that

• Crack specific types of problems, restify library for converting SOAP

• Know your domain

• Engage the community

• Don’t use node for everything, don’t use it for cryto