Since 2005, we've maintained an internal handbook of standards and best practices. In 2012, we opened it up to the world at handbook.imarc.net, with a Creative Commons license (BY/SA). It's a true living document, regularly updated and expanded.
Long, long ago we hosted most of our sites on servers sitting in a closet in our Newburyport office, sharing a T1 Internet connection. Sounds primitive today, but circa 2000, that was pretty much the best available option if you didn't have IBM money to spend.
About a decade ago, we migrated all those sites out to servers in in top-tier colocation facilities in Boston, and or to managed server facilities Chicago andTexas. It cost more to rent racks in a quality hosting facility or have a managed host provider set up boxes than it cost to just use our office T1, but it was worth it: uptime, reliability, and performance all improved.
But that still had us managing boxes, dealing with the occasional drive failure or degraded RAID, and required a trip to the colo whenever we needed to add capacity. And when we were out of rack space, we had to buy another server cabinet's worth of space, which would increase cost out of scale with the revenue supporting it.
We explored managed hosting for a while. Rackspace is every bit as good as claimed, but you pay a hefty premium for their level of service. The value was good, but we're engineers; we don't need that level of support. The price/value didn't make sense for most of our projects. (It does for some. It's just that one size does not fit all.)
A few years ago, we tried out Amazon's services. First we did a little S3 storage, and it was a gateway drug. The next thing we knew we were going wild with the three-letter acronyms: there were EC2, AMI, EBS, EIPs and ASGs everywhere!
The Coming of AWS
We've adopted Amazon's infrastructure for our own. It scales easily and quickly, is far more reliable than colocated servers or a coastal city's power grid, and it has even become cheaper over time, despite every other expense going up.
Amazon Web Services (AWS) is well-known for being the foundation for VC-backed startups such as Instagram, but it works just as well for smaller companies such as ours. Every site we host for our clients now run on our AWS stack.
I say "our" on purpose. AWS provides the CPU, memory and storage resources, but we run our own customized Ubuntu server configuration, updated three times per year, pulling from only the most highly-regarded repos. And we have the flexibility to quickly roll out critical security updates as needed, without breaking the deployment processes.
We've defined that process, and you can learn more about it on the new Servers Handbook page. If iMarc hosts a site for you, now you can see a little more about what goes into managing your server!
Happy reading!
PS: We still love FireHost for clients that need the utmost in security monitoring and rapid response. We like to go home from time to time, but FireHost never sleeps.
