Puppet made server setup and system configuration repeatable with code instead of manual changes. That solved the common problem of machines drifting into different states over time and reduced the effort needed to keep many servers configured the same way.
AWS made infrastructure available as an on-demand service over APIs instead of requiring companies to buy and manage physical hardware first. This solved the big problem of slow provisioning and high upfront infrastructure cost.
TeamCity helped teams automate builds, tests, and later delivery steps in one CI server. This solved the problem of developers building software manually on different machines and discovering integration problems too late.
RabbitMQ gave teams a reliable message broker for sending data between systems asynchronously. This solved the problem of tightly coupled applications by letting services communicate through queues instead of depending on direct, fragile connections.
Google App Engine let developers run web applications on Google’s infrastructure without managing servers directly. This solved the problem of building, deploying, and scaling web apps when teams did not want to handle server operations themselves.
GitHub gave developers a shared place to host Git repositories and collaborate around code. This solved the problem of distributed version control being powerful but still hard to use as a team workflow for reviews, sharing, and open-source collaboration.
Windows Azure gave Microsoft developers a public cloud platform for running applications and using cloud storage, compute, and networking services. This solved the problem of being tied only to on-premises Microsoft infrastructure and tools.
New Relic made application performance monitoring available as a SaaS service. This solved the problem of teams having poor visibility into how their applications behaved in production, especially as systems became more distributed and always online.
Chef let teams describe infrastructure and configuration in code with a developer-friendly approach. This solved the problem of manual server setup and made infrastructure changes easier to version, reuse, and automate across many machines.
Redis provided a very fast in-memory data store for caching, queues, counters, and other low-latency workloads. This solved the problem of traditional databases being too slow for some real-time use cases where speed mattered most.
Heroku made application deployment much simpler by offering a platform where developers could push code and run apps without managing the underlying servers. This solved the problem of deployment being too complex for small teams and individual developers.