PayPal’s Microservices Architecture Journey
<p>During early 2000 the architecture was way simpler than the above diagram. Whenever user interacted with the website a monolithic application written in C++ was handling it. By the year 2004 they already had 76 executables separated, kind of like microservices, but not really. Some services were handling credit card payments, some for eBay payments, etc.</p>
<h1><strong>Problems</strong></h1>
<p>In 2007, it was obvious that this architecture is not going to work. As they still had to deploy everything once a month, when doing a major release. Here is the small list of the problems.</p>
<ul>
<li><strong>Package replication happened on the primary network.</strong><br />
Pushing multi-gigabyte files to machines over the same network that was used for production traffic.</li>
<li><strong>Releases had to happen during off-peak hours.</strong><br />
This didn’t work well once they become a global company.</li>
<li><strong>Required a lot of coordination for a different team.</strong><br />
QA and other teams had to be on a call during the deployment in case something goes wrong. And it took a couple of hours to finish.</li>
</ul>
<p><a href="https://medium.com/@dmosyan/paypals-microservices-architecture-journey-e085452471d0">Click Here</a> </p>