Disaster Recovery Journal Fall 2024

apply the resources, they are often very high paid technical resources which can be more valuable to the company elsewhere. So How Can AI Help to Improve DR Programs? The four main challenges of DR pro grams sound like an ideal problem for AI to address in improving a DR program. Today, AI can address the above-men tioned challenges throughout the DR life cycle. Starting by identifying the IT resources to be protected by the DR plan, using an AI model trained on the cloud resources data for a specific cloud platform (e.g. AWS, Azure or GCP), AI tools can under stand the map of utilized resources and define application boundaries. Based on the results, executable application-level (i.e. not just infrastructure level) DR plans and scripts are generated using AI that can be continuously and automatically tested through disaster simulations. These plans are suitable for both internal usage and external consumption to address any regu latory or audit requirements. In addition to initial discovery and application boundaries, a set of algo rithms to track changes in the resources and applications can be run to discover the changes in the infrastructure resources by identifying the new resources and expired resources, then adding to or deleting from the relevant applications and updating the DR plans and scripts. This process is con tinuously run and tested to ensure the DR plans and scripts of an organization are current and valid. How precise is this algorithm? Today, approximately 90% precision is being achieved. A customer’s technical resource involvement is still needed, but to a lesser degree. The DR process automation includes a continuous testing loop which highlights the remaining issues and helps eliminate them. Almost 100% accuracy is achievable with an accumulation of the infrastructure data across many compa nies. How does the AI model work exactly? Through cloud platform API, a set of fea tures for each cloud resource can be col

lected and then those features serve as input for the AI model. To simplify, all the resources are put into multidimen sional space and clusters of resources (i.e. resources with the closest distance to each other to define application boundaries). The Vision of AI Organizations will be better protected through the utilization of AI to address many of the challenges of DR programs today. Like any new technology, the use of AI in DR programs will grow by incorpo rating end-to-end DR process automation through the mix of deterministic algo rithms and AI models applied continu ously throughout the cloud discovery, DR plan and scripts generation, testing, and tracking changes steps of the DR lifecycle. Executable application-level DR plans and scripts generated automatically in a few clicks and constantly updated reflect ing all infrastructure changes is the new norm for DR. Additionally, DR plans being regularly tested and users getting reports on actual recovery times and recovery points, along with other metrics, for each application should be the new norm using AI. Finally, for those few areas where the AI might need some human assistance, users will get regular reports and use a co pilot to solve any remaining inconsisten cies in the recovery process. As a result of injecting AI into DR, the annual DR drill could be eliminated and replaced with a continuous DR testing routine that reduces business risks by min imizing recovery time and delivers a more assured and predictable recovery. v

 Manual steps – While modern DR tools and solutions have automated many manual tasks in the DR processes, there are still several key processes which include manual steps (plans creation, scripts creation, change management, testing, etc.). All of these manual steps, if missed or not kept current, can cause a DR plan to fail.  Comprehensive DR coverage – What an organization is protecting and to what level is often key to the success of a DR program. An organization can just be backing up their data and taking it offsite, but will allow them to achieve the RTOs set for their organization? Another example is when an organization protects their infrastructure but does not ensure the DR plan takes them beyond that and includes the recovery of the applications and getting the users of organization’s IT resources back using the systems again. The basic question is if the organization is truly protecting themselves with a DR plan that achieves the desired results the organization needs.  Change management – The biggest Achilles heel for any DR program is change and the ability of the organization to capture changes and appropriately update the DR plans and recovery scripts in a timely manner. Most companies’ cloud infrastructure is changing rapidly with product growth and business changes. When considering scale, for most medium-sized companies the infrastructure can be presented by a pool of cloud resources of a few thousand to tens of thousands. This challenge is exasperated when an organization does not have a regular and effective testing program to validate the DR plans and scripts are current, work, and effectively protect the organization as planned.  Lack of resources – Many companies, especially small or mid-sized companies, do not have the technical and business resources to properly, and in a timely manner, update the DR plans and scripts manually and regularly. The task is just too complex, cumbersome, and is very often not even their primary job responsibility. When companies do

Pavel Danilov is a co-founder at Bennudata. com and a serial tech entrepreneur. Pavel was CEO and co-founder of Fridge No More, the leading ultrafast delivery startup in NYC, and most recently served as chief

commercial officer at National Retail Solutions and HYPR. At Bennudata, he has been leading business and product development. Pavel is a proud alumnus of Stanford GSB.

Mark Jameson is a co-founder at Bennudata.com and a board advisor for UbiStor, Inc. With more than 40 years of experience in the IT industry, with the last 20 years focused on data protection and disas

ter recovery with companies like SunGard, nScaled, and Acronis, Jameson collaborates with senior leaders to bring disaster recovery solutions to market and grow adoption.

10 DISASTER RECOVERY JOURNAL | FALL 2024