Do you remember the DDoS attack of 2009?
On July 7th of 2009, there was a DDoS attack on government websites as well as portal and banking sites, which paralyzed their IT services for a while and caused damages worth over $50 billion. Though many people tend to think this was just to paralyze their IT services, the attack actually targeted the infrastructure which operates the services to cause the failure.
Data centers monitor IT equipment around the clock and offer immediate maintenance for system failures, in order to manage infrastructure such as in networks and keep it running without interruption. IT services we see around us are based on this infrastructure control, and this is why effective infrastructure control is crucial for uninterrupted IT services.
Recently big data control started including big data analytics to predict infrastructure failures and respond to them beforehand. Today, I’d like to talk about ITOA (IT Operation Analytics) using big data and its solution.
IT Infrastructure Meets Big Data
So far, infrastructure control has focused on fixing the cause of the problem after the fact. As there are a very small number of people managing lots of servers and systems, it was difficult to find problems before a failure took place.
ITOA, on the other hand, focuses on failure prevention based on big data analytics for their infrastructure management. Take a look at the ITOA key map and see what distinguishes it from the existing control method.
The existing control method was designed to take measures immediately after the fact. It collects data for the infrastructure performance index from the system operation resources, and sets off an alarm to take care of the failure when the collected infrastructure index value exceeds a preset threshold.
ITOA, however, aims to take precautions. It analyses the pattern based on the collected infrastructure performance index, and predicts future infrastructure performance values and patterns. It sets off a preventive alarm according to these predictions in order to take care of the problem before there’s an actual infrastructure failure.
In order to predict failure, ITOA collects infrastructure performance index data for CPU use, memory use, and simultaneous log-ins, then sends it through the big data analytics.
Prediction and Pattern, Keys to ITOA
ITOA using big data analytics can be divided into two types.
One creates predictions on the infrastructure performance index and the other analyses to see if the collected performance index has any abnormal patterns. Both of these methods use the time-series analysis and machine learning which require big data.
Infrastructure performance prediction finds the performance value at a certain time, through time-series analysis. Using this prediction, it sees if there are any performance values which exceed the threshold, and takes precautions beforehand when a failure is expected.
Abnormal pattern detection for performance index learns the normal performance index pattern and analyzes it to see if the new performance pattern is being normal. When the network traffic increases sharply in a short period of time, as it does during DDoS attacks for example, it recognizes the pattern’s abnormality and sets off the preventive alarm as a precaution.
ITOAS Solves Infrastructure Failures
Two different types of big data analytics mentioned earlier are both based on the performance index prediction and pattern detection. The prediction gets more effective when it can learn to analyze more various patterns in the performance index.
The LG CNS data center has ITOAS (IT Operation Analytics Solution), a solution with big data analytics, to perform predictions and pattern analysis for over 50,000 performance index across over 600 servers. It helps you provide services more stably by taking precautions for failure factors.
Today, we learned about how ITOAS predicts infrastructure failures and prevents them by using big data.
Data centers controlling infrastructure are constantly working to make sure their IT services aren’t unexpectedly interrupted. Big data based predictive analytics made it possible to prevent these failures more effectively.
I look forward to seeing more infrastructures with ITOAS for more convenient and stable IT services.
Written by Joon Yeol Yang, Enterprise Solution Analyst at LG CNS Big Data Analytics Consulting Team