Over the last couple of days, the web has been buzzing with problems and solutions to how to stay on-air when a big cloud-service provider such as Amazon Elastic Cloud Computing (EC2) fails to deliver cloud services.
The Tradeshift platform is built on cloud technology including Amazon’s EC2 but was, for multiple reasons, not affected by the outage. The following is a brief explanation of what we do to keep the Tradeshift service running during outages of our own service providers.
First and foremost, Amazon’s outage primarily affected the North American region. Our platform spans several regions including the EU and hence is not dependent on availability of a particular region to operate. In the unlikely (but possible) event that an entire region becomes unavailable, our servers in other regions will respond to service requests. In addition, we use multiple cloud providers to ensure operation in the extreme case that one provider fails entirely.
Secondly, all critical data (customer data, invoices, etc.) is stored in persistent databases with three levels of backup to independent storage providers and physically independent locations to ensure that the platform can be recovered and reconstructed even in the unlikely event of multiple storage service failures. For both service failures and data failures, we have fault-tolerance and recovery mechanisms in place to isolate errors and reliably redirect requests to working servers and unaffected data.
Finally, we monitor our infrastructure continuously both internally and externally and have an on-call team standing by around the clock, in order to bring the Tradeshift service back to our users as fast as possible in the event of an unforeseen failure.
That said we know, we can never plan completely for the unexpected – The Amazon Easter melt-down was a warning of how much can go wrong. And this is where our promise to you, our users and customers kicks-in, we promise to inform you the moment we know something is wrong and keep you up-to-date as we learn more.
