Skip to content

BTW, I’ve been maintaining a reading list on stream data transport and processing:…

(Warning: Kafka-centric. TODO!)

I just backed The Small Silver Speedster: A Father/Daughter/Car Story on @kickstarter

A promising development in the search for treatment for Alzheimer’s disease:…

Idly wondering whether there’s a decent single-player RTS for iPad.

I love this innovative + creative move from @stripe:…

tbh I’ve sort-of got a crush on stripe, if that makes sense 🤔

I’m headed to San Francisco for @apachekafka summit! If anyone wants to hang out I’ve got some time on Wednesday and Sunday, ping me!

Cogent “alternative” perspective on stream data processing

I just re-read this fantastic article: The world beyond batch: Streaming 101 by Tyler Akidau (August 2015) and realized that it provides a super-helpful “alternative” perspective on stream data processing, whereby “alternative” I mean my own perspective to date has been firmly aligned with the folks who created Kafka and Samza while they were at LinkedIn, in particular Martin Kleppmann and Jay Kreps.

Akidau, OTOH, has worked for years on data processing systems at Google, including MillWheel, and Cloud Dataflow, and Apache Beam. I could be wrong but I think his ideas provide a helpful and compatible counterpoint to those of the Kafka-oriented crowd.

I’ve added this to my stream data reading list. I haven’t yet read part 102 but I suspect it might be similarly illuminating.

This explains **so much** of my experience in the world! Holy shit!…

/via @samruby (thank you!)

Apache Beam is really interesting! Seems sort-of like a high-level DSL for stream data processing plus multiple engines for interpreting it.

This is cool!…

#lambda #serverless #docker

Deciding between The Witness and Life is Strange for my next console game — maybe they’re different enough that I can switch between them. 🤔

A small investment to reduce the stress of our car battery dying. Turns out kids love to play with light switches!


.@TDBank_US this is such self-serving bullshit that I’m tempted to close all my accounts after 10+ years:


“The thing is, being connected doesn’t magically enable effective communication.”…


/via @codinghorror

Just finished The Last of Us. Blown away. Such a powerful story. And everything — it works on so many levels.

Tesla Roadster. Great color! @ Pleasantville, New York

Oh yeah, I have a new job; software architect at Park Assist: — it’s going really well!

My AWS Wish List

As part of my new job at Park Assist I recently met with our AWS account representative (along with a few solutions architects). During the conversation I shared with them some enhancements I’d love to see from AWS:

  • Kinesis supporting the Kafka wire protocol would be great because we’re using Kafka in on-site installations (where it’s not possible to use Kinesis) so it’d be very helpful to be able to use the same Kafka libraries and APIs in AWS with Kinesis, so we wouldn’t have to operate Kafka clusters in AWS.
  • Lambda supporting Ruby would be great because we’re adopting Ruby as our primary server-side language, so this would make Lambda more aligned with our stack and in-house skills.
  • A “no-ops” “version” of PostgreSQL in RDS, akin to Aurora would be great because we’d get the excellent feature set of PostgreSQL with the simplified operations and billing of Aurora.
  • A bastion-as-a-service service for VPC would be great because I’ve seen too many teams just set up their security groups to poke all sorts of holes through their VPC, rather than follow the “best practice” of using a bastion host.
  • A pricing model for EC2 similar to Google Cloud’s would be great because it’s just much simpler — rather than having to do capacity planning and forecasting to buy reserved instances, with Google “compute” instances just get cheaper over time the longer they’re active, automatically. This feels very “no-ops” to me.
  • A “serverless” or “no-ops” option for hosting “traditional” Web apps, e.g. Rails, Django, Sinatra, Flask, Jetty, etc. would be amazing because the drastically reduced billing costs and operational costs of “serverless” apps is just fantastic, but there’s still lots of cases wherein we need to run our own Web servers and/or deploy “monolithic” apps. There’s progress being made on this front on top of Lambda and API Gateway, but that’s really/kinda/sorta an elegant hack that’s just a bit too clever to be truly sustainable. This could be an enhancement to Elastic Beanstalk, some new integration mode of Lambda and API Gateway, or a brand-new service I suppose — whatever, I’d love to see this.

If there seems to be a theme here — well, yeah. I’m very bullish on “no-ops” services. Of course there are ops going on — but behind the scenes. Just as there are actual servers behind “serverless” services as well. This seems to me a compelling and critical aspect of, and a logical progression of, the IaaS value proposition.

tags: , , , , ,

It’s 2016 and Mac OS still lets apps steal focus when I’m in the middle of typing 😡

Current status: trying to adopt Continuous Delivery for even v0 of my current project… the only limit is myself

“…more than 325K reservations… corresponds to ~$14 billion in… sales, making this the single biggest one-week launch of any product ever.” 😎

This video of the Tesla Model S being created from raw materials to test track:

Super cool!

I don’t know how to decide between these two nearly-identical health plan options — anyone have any suggestions?


If you use Atom as your code editor and are wondering why the “split pane” commands no longer seem to be working 🤔:…

New York passed mandatory paid family leave! Amazing news!…