Kubernetes
Managing multiple services, ensuring uptime and rapid deployment is a challenge when using traditional system administration. Kubernetes is an orchestration tool that manages the deployments of containers with those goals in mind. With Terraform, Grafana, and Prometheus, they form a suite for deployment and monitoring of services.
We leverage Kubernetes to manage the deployment of services on projects like Sakay.ph. Together with complementary technologies, we are able to ensure uptime scaling with demand. Moreover, we are able increase developer productivity by enabling multiple deployment environments for development and testing, and isolating production, all while providing Continuous Development and Integration. This allows developers to deploy to various environments simply by pushing to Gitlab or executing merge requests.
Linux System Administration
Linux has been the mainstay and de-facto operating system for web servers connected to the internet. There are multiple Linux distributions ranging from the bleeding edge distros like ArchLinux to production-stable ones like Debian Stable. Being free and open-source software (FOSS), Linux has very low startup cost while offering commercial support through multiple providers.
We have been managing Linux servers with various iterations, from vanilla deployment, to scripted ones using Puppet or Ansible, to fully orchestrated ones like Kubernetes. Sakay.ph, for instance, has evolved from a single-machine deployment with a JAR file and NGINX, to a scripted scalable deployment of the routing service using Ansible, to a fully orchestrated microservice system using Kubernetes with CI/CD.
We have proven expertise and tooling for using this data to enable analysis across the mobility network, as well as coordinating with government to quickly respond to disruptions and disseminate information to their stakeholders, empowering them with maps, route editors, and live GPS tracking.
Apache Kafka
Apache Kafka is a persistent queue technology that is used for many applications where data production or processing happens at different speeds, or where an event log needs to be maintained for event-based systems. It’s powerful, but tricky to develop on and manage. For instance, figuring out what keys a message should have (since it may affect partitioning), how to organize topics, how the brokers are configured, how many partitions to allocate, and what is the partitioning scheme are some of the many decisions to think about when using Kafka.
Our experience in Kafka comes from developing data systems around GPS processing for Sakay.ph and some of our clients. Using Kafka, we have been able to robustly process data even when the rate of production and process are different. The system is able to withstand sudden influxes of data and processor restarts.
Google Cloud Platform (GCP)
This is Google’s Cloud offering that has an array of cloud services which cater to various IT needs like object storage (GCS), virtual private servers (GCE), serverless functions (Cloud Functions), Kubernetes hosting (GKE), simple application hosting (GAE), PostgreSQL and MySQL hosting (Cloud SQL), hosted data processing orchestration (Cloud Composer), Big Data analytical engine (BigQuery), etc. While each service has their purpose, the menu may be confusing as many of them seem to do the same thing, which can be challenging for companies looking to adopt Google Cloud services.
We have over 5 years of experience developing solutions on the Google Cloud Platform leveraging its various services. For instance, Sakay.ph
- runs on Google Kubernetes Engine (GKE),
- stores GTFS files and various reports in Google Cloud Storage (GCS),
- uses Cloud SQL for high-availability database with backup,
- uses Cloud Scheduler for triggering scheduled tasks such as managing GKE clusters that runs on Google Cloud functions, and much more.
We can tap our extensive experience to help you navigate the GCP ecosystem with our GCP-certified engineers, get projects safely set up, and maximize bang for your buck.