Trying to manage a sprawling, complex data center via manual data input and monitoring can increase the likelihood of data input errors and can cause IT administrators to miss incidents. But artificial intelligence technology can help admins track all their data center components and automate data collection.
From an operational standpoint, automating operations using AI can lower overall costs and simplify installing software and hardware. Still, organizations must make sure that they implement the right capabilities and have realistic expectations of AI capabilities.
Instead of trying to immediately install functions that attempt to replicate the human brain, organizations should focus on using machine learning and big data for AI. Machine learning enables admins to start building automation on a much smaller scale with individual functions instead of requiring an expensive hardware and software upgrade.
Machine learning is the most common approach to developing an AI framework because machine learning software collects data that is important for future use, integrates it into the function and improves how the machine responds to tasks. Automation just completes the task and moves on to the next task without any data analysis.
With machine learning software routines, data pools and algorithms, admins can train the data center to perform certain tasks automatically, such as filtering spam requests or flagging ransomware attacks. Different types of algorithms include logic programming, decision tree learning, reinforcement learning and Bayesian networks that develop AI capabilities over time.
Investing in machine learning
A machine learning program needs enough intelligence to accurately and consistently identify and take action on current events in the data center, such as distributed denial-of-service and brute-force breach attacks. Being able to automate and analyze event data is an effective way to integrate AI-like capabilities into the data center.
The machine learning software should be able to identify where physical resources will run out and procure and provision new resources in time, as well as shift workloads as required.
A machine learning tool should also report against past incidents and learn from patterns to better optimize itself to deal with every day and future events as time passes to minimize the need for manual admin intervention. Admins utilize these types of capabilities in security information and event management software, but this is only one component of the data center, and admins might want to track data across the entire setup or automate other software products.
Advancing AI capabilities
As AI capabilities continue to advance, the software will become more human-like. The human brain tends to work on probabilities of things happening, but admins tend to add their own perspective and experience.
Ultimately, AI should be able to automatically apply probability factors and remove the need for admins to weigh multiple options. In 2018, systems that do this -- such as IBM's Watson -- are too expensive to implement in an average data center, and few data center administrators would trust any system that can override their professional experience.
Admins should expect to see systems from companies such as Electric Cloud, CA Technologies and HashiCorp that work in tandem with humans, and also set a basis for fully automated AI.
Dig Deeper on Emerging IT workload types
Related Q&A from Clive Longbottom
Hyper-converged infrastructures pose different challenges depending on the deployment method. Software might not offer as much support, while ... Continue Reading
How can you maintain network security beyond the standard firewall and blacklisting tactics? Encryption and digital rights management can ensure ... Continue Reading
SIEM tools deliver automated alert actions, normalize log data and provide intelligent filtering, all of which can help IT administrators lighten ... Continue Reading