The FreeAgent Ops team head off on a two-day road trip this morning as we start the initial phase of our plan to build a new home for FreeAgent, the app.
Since the company was initially founded we have hosted FreeAgent with Rackspace UK, on the outskirts of London. Our infrastructure has grown dramatically over the past 18 months as our customer base has rapidly increased. The introduction of our second generation API, bank feeds, and a need for multiple full-stack integration environments for testing and QA has dramatically increased the need for more servers. Our Puppet dashboard currently shows 66 servers in our cluster, a mixture of virtualised and bare metal, which is spread between managed hosting and the Rackspace Cloud.
As the strength and depth of our team has increased, so have our needs as an Operations team and with it our appetite for moving away from managed and cloud hosting. Not only this, but our needs for maintaining businsess continuity in the event of a serious data centre (DC) outage are increasingly important. We plan for disaster recovery (DR), of course, but we want the best plan we possible can - which, to us, means running FreeAgent concurrently in multiple data centres so our service can continue as normal even in the unlikely event of a full data centre outage.
Which brings us back to our road trip. Over the next two days we’ll be fitting out racks in two ultra-secure data centres with our new hosting partner The Bunker. Once we’re finished, we’ll have 14 Dell R620 servers with a combined 1.7TB RAM and 224 CPU cores (a whopping 448 threads) available to us. We’ll be making full use of Puppet, SmartOS and the network wizardry of our Ops team to provide a state-of-the-art hosting setup which will grow with FreeAgent for the next five years and beyond.
It’s a truly fascinating set up which we’ll be writing more in-depth articles about in the coming months. Stay tuned.