Adventures in Avro

By Tim Perkins on Jun 13, 2016 8:11:57 AM
As part of our microservices architecture we recently adopted Avro as a data serialization format. In the process of incorporating Avro we created a Ruby DSL for defining Avro schemas, developed a gem for generating models from Avro schemas, and reimplemented an Avro schema registry in Rails. Here’s how we got there ... 
Here’s a situation that may be familiar: at Salsify we are moving towards a microservices architecture. Have you done that too? Are you thinking about doing it? This is a pretty common progression for startups that built a monolithic application first, found great market fit and then need to scale both the application and the team. At Salsify, we already have quite a few services running outside of the original monolith, but several months ago we started to define an architecture for how we should chip away at the monolith and structure new core services.
Naturally part of the architecture we are defining is how services should communicate with each other. For synchronous communication, we decided to stick with HTTP REST APIs that speak JSON. For asynchronous communication, we selected Apache Kafka.
We evaluated several data serialization formats to use with Kafka. The main contenders were Apache Avro, JSON, Protocol Buffers and Apache Thrift. For asynchronous communication we wanted a data format that was more compact and faster to process. Asynchronous data may stick around longer and the same message may be processed by multiple systems so version handling was important. A serialization system should also provide additional benefits like validation and strong typing.
At this point, I should inject that we’re primarily a Ruby shop. We love our expressive, dynamic language of choice, so a big factor in the selection process was how well the framework integrates with Ruby. Based on the title of this post, it's not going to be any surprise which option was the winner ...

Read More →

Delayed Job Queue Fairness

By Robert Kaufman on May 23, 2016 1:06:13 PM
At Salsify, we use Delayed Job extensively to handle asynchronous tasks. This works well for us, as it means we can finish web requests faster, resulting in a more responsive web app, while offloading non-urgent tasks to background jobs. For the most part, Delayed Job (and similar job queuing mechanisms like Resque, Celery, etc.) provide a simple and highly effective approach for running background work, making sure that it gets done, and providing a framework to scale your compute resources to handle expected workloads. Even beyond that, these frameworks create straightforward opportunities for dynamically scaling resources to handle spikes in workload. For example, we use an excellent service called HireFire to dynamically scale our Delayed Job worker pools based on queued work. Meaning, we can meet the needs of changing workload while keeping our hosting costs reasonable.

But despite all of the advantages of running background jobs, under real world usage you can still run into challenging situations that require thoughtful handling. One general class of problems that can arise is achieving fairness in resource usage across users. 

Read More →

Efficient Pagination in SQL and ElasticSearch

By Josh Silverman on Apr 20, 2016 10:06:37 AM

Many web interfaces let a user effortlessly page through large sets of data. Implementing database queries that fetch these pages is also effortless for the programmer, usually requiring an OFFSET and LIMIT in the case of SQL and a FROM and SIZE in the case of Elasticsearch. Although this method is easy on the user and programmer, pagination queries of this type have a high hidden cost for SQL, Elasticsearch and other database engines.

At Salsify, we encountered this problem when implementing a feature to allow a user to step through records in a large, heavily filtered and sorted set. We had to implement an efficient pagination solution that would work in both our SQL and ES datastores. In this post we’ll look at an interesting technique to make pagination efficient for the database with large datasets. Specifically, we’ll look at implementation in SQL as well as how to translate this method to Elasticsearch.

Read More →

Good Fences: Neighborly Styling with CSS Modules

By Dan Freeman on Feb 24, 2016 9:57:55 AM

fenceHave you ever noticed that no one writes "How we name our Ruby variables at Company X" blog posts? No one's making the Hacker News front page with "I combined these two strategies for method naming and suddenly my JavaScript is maintainable!" And yet when it comes to CSS, developers are all about naming strategies. We sing the praises of BEM, SMACSS, OOCSS, SUIT, or whatever other set of capital letters is popular this week. Why is that?

Read More →

Using vim/tmux for Ruby on Rails

By Daniel Piet on Nov 25, 2015 8:43:53 AM


Using vim has been one of the greatest productivity boosters for my development life. I got into vim as a lowly system administrator because it seemed to be the tool of the trade. From there, my knowledge grew and now it is my editor of choice for almost all projects. This post will go through my current setup for Ruby on Rails development. I'd like to give a huge shout out to @tpope who has created a plethora of amazing plugins all worth making a part of your daily workflow.

Read More →

In Pursuit of a Scalable Ruby Offline Sort: Adventures in Ruby Memory Management

By Matthew Cross on Nov 5, 2015 8:15:00 AM


Here at Salsify, many of our customers regularly import large amounts of tabular product data into the system. This data needs to be sorted prior to being handled by different parts of the import process. Since we are running on Heroku, memory is a scarce resource. Sorting these arbitrarily large tabular data files requires great care. Reading all of the data into memory at once can result in extremely long execution times due to increased pressure on the Ruby garbage collector and can starve other processes on the same system of memory.

We needed a way to sort large files using a predictable amount of memory. Big data technologies like map reduce were overkill for our data scale. Creating an offline-sort gem to do this turned out to be quite the adventure and forced me to dig deeper into how Ruby manages memory, ultimately requiring a specialized heap implementation.

Read More →

Delayed Job Worker Pooling

By Joel Turkel on Aug 19, 2015 12:05:07 PM

pool-partyAt Salsify we run most of our Ruby on Rails based infrastructure in Heroku. We recently switched our Puma web workers over from Heroku 2X dynos to more powerful Heroku PX dynos and saw dramatic performance improvements: a 51% reduction in mean response time and a 59% reduction in 99th percentile response time. Based on this we did some experiments running our more memory/CPU intensive Delayed Jobs on PX dynos and saw a similarly encouraging 43% reduction in mean job execution time and 66% reduction in 99th percentile job execution time. This was great for a proof of concept, but only one of the eight cores on the PX dyno was being used. In order to take this to production in a cost-effective manner, we had to figure out how to utilize all of the cores on the dyno.

Read More →

Bye Bye STI, Hello Discriminable Model

By Randy Burkes on Jul 24, 2015 11:26:14 AM

At Salsify, we manage the flow of product information between manufacturers, and distributors through to large retail outlets such as Walmart and Google Shopping. Our flexible schema allows users to describe their products however they see fit. We store product attributes in a single table, but each product attribute has a potentially different data type and associated application-specific logic. 

Read More →

Runtime Debug Logging with Ember.js

By Dan Freeman on Jun 26, 2015 6:03:07 PM


When debugging software, particularly if it's code you didn't write and aren't familiar with, one of the hardest parts can be just determining where to start. Without the right context, you can end up chasing red herrings all over before finally tracking down the underlying problem. Well-placed log messages can help provide that context, saving that time and frustration. In this post, we'll look at adapting a logging technique employed by many Node tools on the command line for use in an Ember.js app in the browser.

Read More →

Euthanize Exhausted Worker Processes with a Delayed Job Plugin

By Randy Burkes on Oct 28, 2014 2:18:28 PM


Salsify is a multi-tenant SaaS product information exchange platform hosted in Heroku. We utilize delayed job extensively for long running tasks like import/export/search indexing/etc. One of the features we've grown to love about delayed job is its extensibility via plugins (see our other posts). Recently, an increasing number of job workers have been exceeding their dyno memory quota**, and consequently suffering serious performance degradation. While we are always working to improve our code performance, we thought it would be nice to have a mechanism for euthanizing workers that are unrecoverably exhausted. Enter the DelayedJob::YouthInAsiaPlugin

Read More →