How can AI assist in disaster recovery? That may sound like a simple question, but the answer depends on whether you mean what AI is capable of doing today, versus what it could potentially do in the future.
Previously, we looked at which disaster recovery capabilities are possible using existing AI technology, so we won’t rehash that discussion here. Instead, here we’d like to focus on emerging use cases for AI in disaster recovery – meaning features that don’t really exist at present, at least not in a mature way, but that are likely to become available in the future as data backup and disaster recovery tools make more effective use of AI.
Current state of AI in disaster recovery
Let’s start by covering the basics of how AI can currently be used for disaster recovery.
At present, most use cases for AI in this context remain relatively basic. AI can do things like generate playbooks, analyze data backups and assess systems post-recovery for signs of potential issues. These are certainly valuable examples of leveraging AI to bring speed and efficiency to disaster recovery operations, but they don’t amount to full-scale integration of AI across all aspects of disaster recovery.
It’s worth noting, too, that most AI-based disaster recovery capabilities that exist today require the use of generic AI tools and platforms, like those from OpenAI and Google. That’s because few vendors who develop data backup and disaster recovery tools have added meaningful AI capabilities to their products. Some may claim to offer AI-powered backup or recovery, but in many cases, they’re really just slapping the “AI” label on features that boil down to basic automation or analytics capabilities.
In short, while it’s possible to do some interesting things in the disaster recovery space with AI today, the opportunities are limited, and in most cases the features are not built directly into disaster recovery tools. To quote from an IDC report, “it’s still early days for comprehensive AI in disaster recovery and business continuity solutions.”
Emerging AI technologies in disaster recovery
But just because AI capabilities in disaster recovery are limited today doesn’t mean there is little reason to get excited about the potential of AI in this space. On the contrary, there are plenty of ways in which AI could potentially boost the efficiency and effectiveness of disaster recovery tools in the future.
Following are some key examples of AI-powered disaster recovery features that might be coming down the pike.
AI-driven predictive analytics
One way AI could potentially benefit disaster recovery is making recovery unnecessary in the first place by identifying problems before they happen.
This could be possible using AI-driven predictive analytics features, which would analyze historical data about outages and suggest which types of issues might lead to future incidents. In turn, predictive analytics would allow businesses to mitigate their risks before new outages occur.
As TechTarget puts it, “AI-enabled predictive capabilities can significantly reduce downtime by alerting IT departments to issues before they become critical, allowing them to address problems proactively.”
Sophisticated predictive analytics tools already exist, so disaster recovery vendors likely wouldn’t need to build them from scratch. Instead, making a feature like this work in practice would mainly require enough data about historical outages trends to enable AI tools to identify meaningful patterns and provide relevant guidance to help IT teams mitigate issues proactively.
Self-healing systems
If, despite an IT team’s best efforts to react to warnings about potential issues, a system still fails, AI could also help in recovery by automatically “healing” the issue. For instance, if an application crashes, AI could determine why it crashed, then resolve the issue, without requiring explicit guidance from humans. The result would be faster recovery with lower effort on the part of engineers.
The notion of AI-powered self-healing systems is not especially new; it has been discussed for years in the context of AIOps. But most discussions have focused on the theoretical, not the practical, because to date, the ability of AI tools to resolve problems on their own has been limited. AI might be able to fix simple issues, like broken network settings that prevent successful copying of data from backup storage to a production system during recovery operations. But AI in its current form is less likely to be able to do something like fix a memory leak bug inside an application’s source code, then recompile and redeploy the app.
That said, it’s not impossible to envision use cases like this. AI is already capable of generating code, and processes like compiling and deploying apps are easy to automate. It’s just a matter of linking capabilities together in a way that enables true AI-driven self-healing.
Setting recovery priorities
In most businesses, some IT assets are more important to operations than others. This means that in the event of an outage, teams will ideally recover the most mission-critical systems first, then move onto restoring less important ones.
Currently, an organization’s ability to decide what to prioritize is a manual affair. Engineers must interface with other stakeholders within the business to understand which systems matter most from a business perspective. They must also take into account the technical requirements of each system and the complexity of restoring it following an outage. And they need to translate all of this into a disaster recovery plan that makes it easy for technicians to decide what to prioritize.
With AI, however, the process of setting priorities could become much more automated. Instead of manually collecting information from business users about the importance of each system, AI-powered chatbots could collect the data, then assess it alongside technical information to generate disaster recovery plans accordingly.
A capability like this would require a blend of analytical AI and generative AI technology. This type of feature doesn’t currently exist, but it’s not hard to imagine implementing it.
Intelligent Backup and Restore
Along similar lines, AI tools have the potential to formulate more intelligent backup and restore strategies. By “intelligent,” we mean plans that are designed to be as efficient as possible by avoiding redundancies or wasted time.
For example, a business might currently operate two instances of the same database in order to increase the reliability of the database. The data inside the two database instances is identical, so the database really only needs to be backed up and (in the event of an outage) recovered once in order to restore normal operations. But because the organization’s backup and recovery tools are configured to back up and restore all assets, each database instance is handled separately. This leads to redundancies in backup data. It could also slow down recovery because the same database would effectively be restored twice.
AI tools designed to identify redundancies in backup and recovery strategies could flag an issue like this as a source of inefficiency. They could even potentially update backup and recovery configurations automatically to mitigate the issue.
Here again, the AI capabilities necessary to build a feature like this exist in the form of sophisticated analytics tools. But to enable those tools to help optimize backup and restore strategies, the algorithms that power the tools would need to be customized to recognize trends related to data backup and recovery. They might also need to be trained using data sets that reflect “good” backup and recovery configurations. This can all be done easily enough, but it hasn’t been done yet.
Automated compliance monitoring
Today, ensuring that backups meet whichever compliance obligations a business faces is a largely manual affair. It involves demonstrating to auditors that backups are in place, and that they adequately address compliance risks.
With AI, however, this process has the potential to become much more automated. Compliance requirements could be defined using code, and AI tools could automatically verify that the requirements are met.
An approach like this would not only save time on the part of auditors. It would also reduce the burden placed on IT staff to supply compliance information. And it would add consistency to compliance reviews because the compliance process would be automated, eliminating the inconsistencies that can arise when one human auditor interprets compliance rules or evidence differently from another.
Compliance as code is not a new idea. However, few tools or frameworks currently exist for expressing compliance mandates using code. Nor are there AI tools designed to interpret code-based compliance rules and monitor systems in response to them. But none of these types of solutions require magic to work. It’s feasible enough to imagine vendors implementing them.
Don’t wait on AI: Streamline disaster recovery today with N2W
While AI-powered features may offer exciting possibilities for the future, N2W is committed to delivering sophisticated data backup and recovery capabilities that businesses need right now to ensure rapid and reliable disaster recovery. The tool’s cloud-native IaaS approach provides a secure, fast and streamlined backup and recovery solution that allows the user to maintain control of their data within their own environment.
By choosing N2W today, organizations can:
- Leverage proven automation features (such as DR drills and reporting) without waiting for AI advancements
- Recover data across multiple cloud regions, accounts, and public cloud providers
- Back up critical cloud network settings alongside core workloads
Sign up for a free trial to see for yourself how N2W can streamline your disaster recovery strategy.
Tomorrow’s technology isn’t protecting your business today and as data losses, outages and ransomware attacks increase, strategies must be in place…yesterday.
