Services
Grimes IT provides technical services to web-based businesses.
Full stack web development
Grimes IT specializes in building web applications–software applications that run on the World Wide Web. Full stack development means support for every technology layer used by the application, including user interfaces, data, business logic, APIs, infrastructure, development and deployment.
User interfaces
In modern web development, user interface often means JavaScript.
I use client-side JavaScript libraries like React, Vue, and Angular to provide rich user experiences. Additional “batteries included” frameworks like NextJS (for React) and NuxtJS (for Vue) can be helpful to add some server-side magic, and to simplify best practice and architecture decisions.
Some cases still call for old school SSR (server-side rendered) PHP applications. I’ve built a generation of PHP apps with a variety of architectures, including full-stack frameworks like Symfony, Zend/Laminas, and Laravel; microframeworks like Slim and Silex; home-grown frameworks; and third-party ecosystems like Wordpress and Moodle.
A good user interface needs a good designer, who is often an outside specialist unfamiliar with the application’s underlying technology. The architecture should make it as convenient as possible for outside designers to contribute to the user interface.
Databases
Data is the core of a web business, and it needs to be stored and used in a secure and efficient way.
Relational data stores like Postgres or MySQL are often a good choice for the core database, though NoSQL options like MongoDB can have their place. I spent many years pushing MySQL and Postgres to their early limits to scale web apps. I like common ORM libraries for basic database interactions, and I’m experienced at using complex (yet performant) SQL to work with the more interesting cross-sections of data.
Atop the core business data, many other technologies can be stacked in various combinations, to leverage or compile data for a variety of specialized purposes. Some common ones I’ve used include Redis, Elasticsearch, Memcache, OpenAI, etc.
Business logic and APIs
Business logic is often distributed across a variety of applications and third-party services, which all need to be integrated.
Often fundamental business logic is embedded in an original web application, and it needs to be extracted in order to be more easily shared, tested, and refined. This means APIs.
REST has become one of the most common API architectures, and I’ve built and consumed many REST APIs over the years. I also like GraphQL for solving some of the shortcomings of resource-based APIs.
Web infrastructure
From bare metal to cloud, from server to VM to container to serverless, I’ve scaled real-world systems for decades. Before that, I was a network engineer in Silicon Valley. For me the “full stack” includes the OSI networking model.
Development environments
I run all my development environments in Docker containers now, so I can install whatever mad dependency a project might require without messing up my own computer. Containers are also useful because they can closely reproduce entire production environments (databases, caches, cloud services, and all) and even simplify deployment, infrastructure, and operations.
Visual Studio Code and Docker Desktop have made it easier to do development this way.
For new projects and quick changes, GitHub Codespaces can be an easy way to get a simple dev environment up and running, and to make quick changes on the fly, without having to hassle with creating a local dev environment first.
Production infrastructure
Production infrastructure for web apps is sold á la carte in a range of service levels with the general idea being the more it costs, the less work you have to do yourself, and vice versa. Every business has to find its own sweet spot for its own needs and budget.
For new or experimental applications, full-featured managed platform providers like Vercel, Heroku, or Digital Ocean can be useful. But they tend to become overly expensive beyond early stages.
Managed service providers can be valuable for specializing in particularly complex or important parts of the stack, like databases and network services. Just beware of vendor lock-in.
For everything else, infrastructure services like AWS tend to be the most practical and economical choice. Consider that most managed service providers are probably using AWS or something comparable themselves, so you’ll pay them whatever you’d pay AWS anyway, plus the additional markup for their services. Some will be worth it. Some will not.
Bare metal servers can still be run in this day and age, especially if you enjoy that sort of thing. But for me it’s only worth the trouble when an application’s resource usage doesn’t match the horizontally scalable models of the cloud providers.
Deployment and DevOps
Deploying a change from development to production should be easy and safe. Pre-release testing and post-release monitoring are two sides of the same coin, and both should happen automatically.
I like GitHub Actions for automating deployment workflows and I’ve used a variety of build environments and CI/CD toolchains.
Modernizing legacy applications
Respect the legacy. Software only survives to become old and messy because it’s been providing genuine value to the business for a long time.
Successful applications endure, and technical debt accumulates.
Getting the legacy codebase unstuck
A poorly documented legacy codebase is best approached by creating good documentation. That gets everyone literally on the same page about important parts of the app and business, and it’s a valuable work product to deliver the client early in a project.
Since legacy codebases can be fragile and risky to change, basic testing and monitoring can make changes safer, detecting problems before or after deployment, and making it easy to revert if needed.
Any new architecture should be created in parallel, while the legacy application keeps running for as long as it’s needed. This requires the architecture to be split somehow, to allow simultaneous work on both the old and new. The “Strangler Fig” design pattern is a useful way to think about this challenge.
Code review and training keeps the legacy codebase gradually improving, along with the team that maintains it.
Grimes IT provides code review as a service and customized training.
Integrating new technologies
Having an established legacy application doesn’t mean you can’t have shiny new applications, too.
To integrate new technologies, while keeping all the useful features of legacy services, it helps to encapsulate the legacy system into a “black box” with convenient APIs, to abstract away the internal complexity of the legacy codebase from other projects.
The legacy API can then be exposed using a variety of transports and architectures (HTTP, Websockets, REST, GraphQL, etc.), as newer technologies require.
Upgrading ancient LAMP stacks
According to 2023 surveys by W3Techs, around 75% of websites use PHP, and around 20% of those are still running PHP 5.
PHP 5 has long been officially dead (“end of life”), and is well known to be at least twice as slow as newer versions. Sites don’t stay on PHP 5 because they like it. Upgrading a core technology like PHP, MySQL, or Linux that has been powering a real business for a long time can be challenging and disruptive.
But it can be done. It takes a well-thought-out migration plan, practical testing and monitoring, mentorship and training, and some patience.
Estimate your project
Does your business need help with something like this? Contact me.