Based on what we built at KickApps, the infrastructure built for web development and as a production environment is designed to scale and not fail.
KickApps, at its highest build-out of infrastructure, was capable of a peak load of 86 Billion dynamic hits per month; that’s a huge number. This reflects the computationally hard requests requiring processing before responding, and doesn’t include the static hits for easy stuff like images, audio and video.
The pattern followed below mirrors the KickApps method in that databases and file systems are replicated in real-time to ensure a system loss will not compromise the service. It also allows for maintenance to be performed on half of the infrastructure at a time without disrupting what matters most; the running applications.
While we’re hosting neutral, we’re happy to facilitate solutions that are smart:
- scale appropriately to your needs
- cost effective
But most of all, we’re happy to talk with you to ensure we and you understand your needs and set you on a path to success; we want to ensure we both reflect fondly on our shared success stories so we can have more of them.
We provide numerous hosting options and are completely flexible as long as the solution works; e.g., it must scale and provide reliable up time that’s appropriate to your business at a cost’s appropriate.
That can be a well known provider, on your own systems or on our infrastructure.
About Our Web Infrastructure:
Reliability and performance is extremely important. Our infrastructure is overbuilt for performance and data, file and application redundancy.
Our web environment hosted behind two highly available (redundant) load balancers. The two production servers have eight core processors, 32GB Ram and two SSD drives and run Ubuntu 16.04.01 LTS. They each run MySQL running Master/Master replication. The file system is replicated using GlusterFS. We can sustain the loss of half of the infrastructure while remaining 100% operational. The development servers are similar in theme but smaller in capacity.
The infrastructure’s applications and database schemas are backed up twice daily to a NAS. Monitoring is done via Zabbix.
We provide this infrastructure as an option as it provides the computational power equivalent to what would lease for ~$800/month from a data center provider at a tiny fraction of that cost.
All systems and networking equipment are connected to four discrete Uninterruptible Power Supplies to sustain minor disruptions. It is not as resilient as data centers with multiple connections to the internet and diesel generators.