The differences between “Hot”, “Warm” and “Cold” Disaster Recovery (DR)08 Apr 2018, Posted by Azure in
With the widespread use of Public Clouds, concepts like High Availability, Data Durability and Infrastructure Elasticity became an affordable reality for any company, of any size, and not anymore the concern and exclusive domain of large corporations.
Being made of sets of data centers established in different regions, the Public Cloud offers its shared infrastructure to companies looking for systems available 7×24 “no matter what”, since there will be always regions available, even if one (or more) is down.
In this brief article, we will explore how Public Clouds meet the requirements for IT Business Continuity and describe the three general types of Disaster Recovery using such global shared infrastructure.
Let’s first bring two indicators associated with IT Business Continuity and Disaster Recovery: Recovery Time Objective (RTO), which is maximum acceptable time services can be unavailable for users and Recovery Point Objective (RPO), which is the maximum acceptable amount of time data can be outdated when a major failure happens.
The most basic form of Disaster Recovery is called Cold, because there are no active (“hot”) services running at a different location (Disaster Recovery site) and all services are provided from a single location or data center.
Periodic backups of the active location are made at intervals defined according to RPO, be it a day, an hour or any other objective. When a major failure occurs, a backup, previously copied to a different data center, will be used to restore operations at the Disaster Recovery facility, which typically takes up to a few hours (RTO). When the original location returns to normal operations the process is reversed, usually causing some additional disruption for users. This reversal is usually done to keeps costs down.
It is important to note that having periodic backups is only part of the Cold Disaster Recovery process, while many other aspects must be planned, including, but not limited to, servers activation, networks, security, domain name resolution, reverse backup, etc. The picture below illustrates a Cold Disaster Recovery scenario.
In the other extreme, there is Hot Disaster Recovery, which provides no down-time (zero RTO) and updated data everywhere (zero RPO). This requires services and databases to be distributed and synchronized across different data centers, usually geographically isolated, in order to preserve operations continuity in case of a major failure in single (or even multiple) data center. The picture below illustrates a Hot Disaster Recovery scenario.
Public Cloud providers offer not only geographically dispersed data centers but also the mechanisms to synchronize data and distribute users requests to the closest site, as well as distribute the load among the active data centers running those systems, in order to avoid excess capacity, as Cloud costs are variable and accrued according to actual usage.
In between Cold and Hot Disaster Recovery there is a continuum of intermediate RTO/RPO requirements called Warm Disaster Recovery. These setups can be closer to either extreme, Cold or Hot, according to Business requirements and available budget.
Disaster Recovery requires extensive planning, periodic reviews and tests to work as designed. Companies should first determine their business requirements in terms of IT continuity (RTO/RPO) and plan to meet these goals in the most cost effective way. CloudNow can help you plan for Disaster Recovery to meet your goals and budget, independent of Public Cloud vendors.
Although this article was written having in mind the use of Public Clouds to provide Business Continuity, the concept of Disaster Recovery is older than that and can be implemented using private data centers or a combination of private data centers and Public Clouds, in a hybrid environment.