I have several customers for whom I maintain Linux servers providing small-scale web hosting, simply because they provide services such as application management, SEO and web design, where the customer will assume that they are also able to provide web hosting.
Generally the setup is simple – a single rented server running one of the popular web hosting control panels – Plesk, cPanel, VirtualMin and the like. Having managed web hosting services in various capacities for 15 years, I believe this is a perfectly acceptable setup, but it has a few flaws that will catch out the unwary:
- DNS: often a single dedicated server is used as the only nameserver for a domain, meaning that if the server is down, it can cause issues with email bouncing, Google removing results, and customers seeing confusing error messages.
- Backups: it can be tricky to set up and test proper off-site backups, especially once the volume of data begins to grow – keeping a few weeks worth of daily offsite backups to another provider in another city will protect you against issues with hardware, software, users, and your server provider.
- Email backup: Internet email has a built-in capability to handle unavailable servers – “backup mail exchanger” or “backup MX”. In a single-server environment this is often overlooked, both from the view of hosting customers who have their own in-house email server, and a backup system able to receive email for your customers when your server is down.
- Replication: a busy e-commerce site taking orders every few minutes, where you rely on the website itself to track orders, would result in a large amount of administrative work for you and your customer should the worst happen
- Security: making sure the server is properly configured to block unwanted access is vital, and although most servers now come pre-configured in a secure manner, there are still things that should be checked and changed with most providers.
- PCI Compliance: for e-commerce websites especially, PCI compliance takes security and reliability a step further, generally involving scanning/monitoring of the server by a 3rd party. This can highlight subtle security and reliability issues that might otherwise be missed.
- Monitoring: there are many website monitoring services out there, and many companies that still do not monitor their websites. It’s generally enough to monitor one or a few websites on a given server – problems that show up in monitoring are generally DNS, server capacity, or database related, so just monitor the main website(s) on the server.
These are the background issues that don’t usually get thought about until disaster strikes, so I’m pleased to say that so far, all the customers who currently consult me on the above have yet to have a major disaster.
So this idea is simple: take the above services, find or create a provider of each service, integrate them all together into an easy-to-use service, and integrate that with all the popular control panels.
I’m currently at the research stage with this – finding out what people would add to the above list, and what sort of price they would be willing to pay. Please add your comments and thoughts on that, and when I have some time I’ll set the whole thing up.
In the meantime, feel free to ask for advice on any of the above – I can recommend providers of services that help with each of the above areas.