#ZertoCon Recap: 6 Pragmatic Rules for DR Implementation
As you may have read in my previous blog, I recently attended #ZertoCon to discuss the changing landscape of disaster recovery (DR). One thing I can say for sure is DR today looks nothing like it used to (spoiler: it’s easy!) and has become a non-negotiable requirement for virtually every organization.
But it hasn’t always been the norm. Let’s take a quick trip down DR history lane:
- “Computer centre managers” and their mainframe’s become a key part of business processes and the recognition of the need for DR became widespread
- The PC revolution – backups become an essential ingredient to DR
- First commercial DR solutions start operations
- Elemental regulations are published
- Client/Server and off-site tape backup solutions are the norm
- The Internet era, backups everywhere, commercial DR solutions the norm
- Virtualization, Cloud and the emergence of offsite replication
- 99.95%+ Uptime becomes the primary goal
One of the biggest catalysts driving DR is the increasing business complexity that’s led to a greater reliance on IT teams to keep their enterprises running at peak efficiency. Also, cloud computing has now unlocked a level of DR preparedness that had historically been financially infeasible for most. The cloud has fundamentally rewritten the rules for DR providing greater levels of protection, security, and flexibility at far lower costs.
As a result, over the last several years the expectations and market demand for DR have changed. The Disaster Recovery as a Service (DRaaS) market is estimated to grow from $1.4 billion in 2015 to $11.9 billion in 2020, at a compound annual growth rate (CAGR) of 52.9% from 2015 to 2020. In regional segmentation, North America is expected to be the largest market in terms of market size while Latin America, Asia-Pacific and the Middle East and Africa are expected to emerge rapidly in this market. Over the last several years the DRaaS market has matured greatly, with more diversified product offerings and lower monthly service costs, leading to a wider appetite for DRaaS across organizations of all sizes.
Often when we talk about DR, what comes to mind is a scene straight of a Michael Bay blockbuster:
Yet, in reality, most IT disasters are caused by things much more commonplace, such as this:
According to Gartner research, the causes of unplanned downtime are not often fire and explosions but environmental factors and power disasters (20%), operation errors (40%) and application failure (40%). Regardless of the cause, any serious disruption to daily business operations is going to be unwelcome, but with the right DR plan in place the impact can be minimized.
To ensure IT infrastructure can handle any type of disruption businesses should follow these six pragmatic rules for good DR implementation:
Get close to the application: The higher in the application stack, the less data that gets replicated, therefore optimal replication occurs as high in the stack as possible. Additionally, managing replication at the application layer is often the only way to achieve transactional consistency.
Latency, not bandwidth: High bandwidth networks have become commonplace, and connections in the 1G or even 10G range have become affordable. One must be sure, however, to understand the impact of latency on replication traffic. Look for a target site that is close in network terms (latency), yet one that offers diverse power and connectivity from the target site.
One size doesn’t fit all: Data have varying recovery point objectives (RPO), therefore applications should be grouped into tiers based on criticality. Also, replication tools have strengths and weaknesses for different use cases so it’s important to tailor solutions for complex environments.
Don’t change form factors: Changes to form factor (physical, virtual, cloud) may result in changes to performance or scaling. Disasters are not the time to experiment with architectures. If an application can be virtualized, do so in production first, not as part of the DR strategy.
Management models don’t mix: Workloads should be managed by the same entity for both source and target. Hosted applications managed at the same service provider for primary and secondary sites are much easier to manage than a mixed model of enterprise IT managed on-site and a service provider managed target.
Backup ≠ DR: Many businesses make the common mistake to assume that backup equals DR but that is truly not the case. While some tools can be used for both purposes, they generally can’t do both efficiently. Backup tools do not typically yield the RPOs or RTOs required for modern business operations, and DR tools often don’t meet long term archival retention needs.
DR isn’t a one-and-done exercise. It’s an ongoing, continuous process, a discipline that requires careful and regular maintenance. Yes, it can seem like a hefty task but luckily, with the advancement of cloud-based DR services and solutions, organizations can get the automation and flexibility they need to keep operations running as usual with less cost and time investment than ever before.
Learn how to take advantage of all the benefits cloud-based DR delivers in this white paper, How Cloud Services are Transforming BCDR.