What do you get excited about? If you have a life, the answer probably isn’t machine learning. It’s arcane, sciency and downright neckbeard-y. Worst of all, it can be hard to see why you should care. There is no 20 minute tech talk you can watch, drop into your favorite editor and be on your way to winning the Netflix Prize. The value proposition can be a tough sell.
As it turns out, it’s pretty bloody exciting to me. If you’re still here, maybe I’ve got a shot at convincing you. Read on, brave adventurer!
We’ve mentioned it before, but we’re currently in love with breaking functionality out of our monstrous core app into smaller, more manageable service apps. There’s still a lot going on in PipelineDeals City, but some nice suburbs are popping up that offer all sorts of amenities. (You simply can’t beat the freshness of the farmers' market in Calendar Sync Township. Starting to wish that’s how we actually named these things.)
Life has been pretty great living on the outskirts. Domain knowledge is nicely contained and kept safe from the high crime rates and outrageous rents we were paying in the big city. But every now and then, we start to miss the convenience of what we once had. As crazy as it seems, it feels like life was simpler back then. But that just can’t be.
I want to take a look at some of the problems we’re facing in this new world. We’ve overcome them all in temporary ways, but let’s ignore those solutions for now. Let’s figure out together how to make them better before we get lost for good. (These new towns always take forever to show up on the map.)
At PipelineDeals, we follow the microservice architecture pattern. Many of our features are separate applications that expose a REST API. But this poses a challenge for testing our service applications.
This post describes a strategy for using an adapter to isolate the service in question, and then outlines a different strategies for testing the service integration.
Well that made people angry.
But amidst all the hoopla, some people did make some really good points about me coming across as a childish nincompoop. Also, I sounded too condescending and derogatory of the incredibly valuable field of devops, and didn’t judge fairly. All fair accusations.
This didn’t upset me. I mean, such a dousing of founded criticism is always a little hard to stomach. But it didn’t upset me a lot, not really. Maybe if I had written the article recently it would have. But I wrote it a year ago! So I sort of watched this all roll in from a distance. People were displeased with past me! But hey, so am I.
However, I am not displeased with my decision to stop using Vagrant. I think that was definitely the correct decision for our team, at the beginning of 2014. I think I wasted a lot of time in 2013, trying to keep our Vagrant hobbling along.
This coming from the only person on our team to ever push for Vagrant! I championed it for years. I still think of it fondly. (I certainly overstated the title of the last article. I tried to make the whole thing quite sensational. Perhaps people reacted against my sensationalism. Fair enough.)
Maybe someday we’ll return to it. But not now. And here’s why.
This was originally published on Chad’s personal blog on 21 Feb 2014.
Hey, kid. Someday, some charmer is going to smooth talk you.
Why would you want MySQL and PostgreSQL both running on your machine all the time? they’ll probably ask.
Just run them when you need them.
Then they’ll tell you about some slick way their company allowed all of their developers to click buttons on some website, like ordering a Fatty Meal at a McDougle’s:
MySQL, Redis, a custom queue server, and our main app, please. Then BOOM. They get a fully configured VM, just like that! With exactly the things they need and none of that other crap. They spin it up, they get their work done. No fuss, no hassle. No having developers waste time configuring their environments.
And most of all, your charmer will woo you with how you’ll never again have developers saying,
it works on my machine! That’ll be their biggest selling point. It will also be their biggest lie.
At PipelineDeals, we have a growing number of
satellite apps—we still have one main monolithic Rails app (which we’ve started calling
p.core), but ever so slowly, we have ported bits & pieces of functionality out to smaller, separate apps. These apps communicate with p.core and each other in a couple of ways:
Essentially, our main Rails app (p.core) broadcasts events that happen in the system to any satellite app that pays attention. So, for instance, if a person’s email address was updated, we would send a pubsub message like
person:updated:123456 and the payload would include information about the attributes that changed.
Conversely, when a satellite app needs to make a change to a model in p.core, the satellite app will use our public api to write changes.
When using our own APIs, we’ve grown into a pattern for wrapping them with Ruby classes that feels very clean and predictable to us.
We on the PipelineDeals engineering team feel proud of the work we do! But you wouldn’t know it by looking at the previous (lack of) content on this blog. We aim to change that.
So why the name change? Reason 1: our previous WebTwoPointOh-style name for it,
PipelineDeals DevBlog, makes us want to take a nap. Reason 2: software engineering has a lot in common with plumbing (an idea I stole from @ngzax, who probably stole it from somewhere else).
At PipelineDeals, we deploy code frequently, usually 2-3x per week, and sometimes even more often. As all web application developers know, deploying is sort of a nervous process. I mean, sure, 99.99% of the time, everything will go perfectly smooth. All your tests pass, your deploy to staging went perfectly, all the tickets have been verified. There is no reason to fear hitting the button. And, the vast majority of the time, this is true.
But all web application developers also know that sometimes, there is a snag. Sometimes the fates are against you, and for whatever reason, something goes bust. Perhaps you have a series of sequenced events that must occur to deploy, and one of the events silently failed because the volume that /tmp is mounted on temporarily had a 100% full disk. Perhaps that upload of the new assets to S3 did not work. Perhaps you did not deploy to ALL the servers you needed to deploy to.
And then, the worst happens. For a short period while you are scrambling to revert, your customers see your mistake. They start questioning the reliability of your system. Your mistake (and it is yours, even if some bug in some server caused the problem) is clearly visible to your customers, your bosses, and your peers.
PipelineDeals is hosted on Amazon’s AWS cloud platform, and has been since 2007. During these years we have been exposed to two separate mass outages, all of which affected their US-EAST availability zone.
Compared to other AWS availability zones, US-EAST center is their busiest, has the cheapest hourly server rates, and (happens to be) the most prone to massive outages.
Given that the majority of our customer base happens to be closer to the east coast, we keep our servers hosted in US-EAST. What this means, however, is that we must be prepared to jump ship to another availibility zone with as little downtime as possible.
The developers at PipelineDeals are happy to announce a new version of our API. V3 introduces many changes, including the following: