Most business & IT folks worth their salt know that backing up their important data is a smart business decision. But all too often we hear nightmares about when a lack of planning results in a critical business failure. Having been in the web hosting business for over 17 years we can tell you that if we didn’t have a solid data backup strategy in place then quite honestly we just wouldn’t be around long enough to write this post.
We backup our Shared Hosting & VPS servers every night and keep a rolling history of 90+ days so that if a customer has an issue from say 48 days previous we can easily pinpoint that exact data and do a restore for them with surgical precision. Our redundant Canadian data network consisting of three nodes in Canada (Vancouver, Halifax, & Calgary) enables us to provide truly geo-redundant Cloud Storage data backup solutions for our internal systems as well as for our clients with dedicated servers and highly available Canadian Cloud IaaS. Based on our experience, here are some of the basics to consider when doing your disaster recovery planning (DRP). There are 3 main types of Cloud Data Backups:
Cold Restore Site
This solution is the most cost effective of the three and, as a result, is the most common. Essentially your data is backed up (typically nightly or weekly) to a remote data center that is ready to be restored in the event that you may need it. If disaster strikes, our engineers will help you get your systems back online with minimal delay. Typically restore times are defined in hours or days.
Hot Restore Site (High availability)
Primarily due to the cost of essentially doubling your infrastructure, this strategy is rare but certainly considered to be the most comprehensive. Hot sites are basically live mirrors of your infrastructure. The redundant backup site is populated with all the servers, networking details, and all of your data is replicated in real time between the two environments. This syncing allows for minimal (potentially zero) impact or downtime to business operations in the event of a significant outage event to your main datacenter.
Warm Restore Site (Cloud Backup & Restore)
With slightly higher costs than with a cold storage approach and with most of the benefits of a hot restore site, this is the most common approach implemented in modern virtualized (cloud) server environments. Warm sites offer datacenter space with pre-installed hardware, virtual environments, networking capability that are at rest until needed. Data is sent to the site nightly (or in some cases hourly) and is made easily available for restoration in the remote environment Typically SLA’s around restore times are set in hours or even minutes. With the advent of cloud technology this approach has become the clear winner. But beware, few providers have a complete and comprehensive solution to do this very well today. Most are just repackaging up old cold storage solutions but there are some key differences to note to look for with a true warm restore DR solution:
- You should not have to pause, suspend, or otherwise interrupt in any way the server as it’s being backed up.
- Having a comprehensive plan for dynamically managing the networking stack including a redundant DNS platform is critically important (otherwise it’s just a simple backup & restore system).
- Live data is tricky (especially with databases), snapshots must be “Database aware” to work well when you really are counting on them.
- A normal modern sized server should be able to be restored in about 2 hours or less.
- You should have visibility into the reporting AND the ability to restore on demand via a user friendly web based management console.
Some Key Best Practices To Consider:
- Local vs. Remote Storage If your provider only offers local backup options then you are at risk in the event of a site wide outage (think New Jersey in late 2013).
- Always use Encryption! If your replication platform or provider doesn’t provide you with encryption by default then it’s just quite simply not an effective modern solution.
- The Cloud is a Game Changer for DR Modern platforms provide software defined networking (SDN) meaning you can setup routing, so that there can be nearly no downtime can almost zero even with a warm site recovery strategy.