Applying IT Analytics to IT Operations Data
IT analytics comes in a lot of different forms. The recent report from Forrester on IT analytics,Turn Big Data Inwards, shows some examples of the types of algorithms that can be applied to IT data.
- Economic algorithms: look at data from a market based perspective, examining supply and demand factors to determine optimum utilization of resources
- Statistical analytics tools: Infers the existence of relationships where explicit (topological) relations are either weak or missing. Statistically compares performance patterns to identify common behaviors and therefore, implicit relationships.
- Configuration tools: look at configuration parameters and how things are set up and built, making sure that there is consistency and looking at how the configuration changes over time.
- Topological tools: Maps relationships among the involved components to narrow the scope of processing and to evaluate upstream and downstream impact.
Configuration Analytics: The Bottom Line for Good Change ManagementChange is really the derivative of the configuration. As configuration changes over time, the changes that occur will be changes in the configuration state over time. As configuration state changes, IT analytics can present where something changed and show the scope of the change. Any time something changes with the configuration, this must be dealt with from a change management perspective. When IT operations understands the configuration state over time, taking into account ongoing changes, then IT operations can also identify when change occurred, to be more proactive about how to plan for making changes.
Identifying When Changes Have Occurred
IT operations has come to understand that when incidents go off the charts and the service desk is busy answering the phones because everything's gone crazy, the first question asked is 'what changed'? They look into the infrastructure, where IT is always monitoring, and checking whether the network is down or the server went bad or the disk drive failed. Nowadays, with a more stable infrastructure, incidents come out of having made changes in the wrong way. By tracking some of the metrics and a key performance indicator, like something related to the applications performance, IT operations can take this as a statistical baseline and realize when something has exceeded the baseline, and know that something happened. Now they just have to understand why.
By looking at the various configuration items and the various configuration parameters and attributes, IT operations can do a pretty good job at understanding the possible cause. When IT operations has 1000s of millions of attributes and 1000s of different components located around the world, this can't be managed manually. IT operations needs to apply analysis to be preemptive and make sure they're not making the wrong changes.
"I fly a lot so I am very happy that Boeing and Airbus and all the others build their planes using good engineering principles because if they built their planes like we build IT, I would have been dead a long time ago. So where we're looking here is at the services that we provide, we really want to apply some good engineering practices."
Glenn O'Donnell, Forrester
Good Engineering Demands Good Visibility
IT operations needs to understand all the different parts of IT environments and how those parts come together. By being open to failure, IT operations can learn from the failure in order to anticipate failure, using science and good engineering to avoid failure. By planning ahead, IT operations can design for that failure so that it doesn't occur.
There is a siloed thinking that doesn't understand how the server, the network, the storage system or applications all come together. IT needs to go through a design and development process and approach IT operations on a more comprehensive level, using a lot of models and playing some 'what if' games. What if this happens? What if that happens? How is that going to impact the outcome?
"This is very analytical and even though we have a lot of people in IT that call themselves engineers, I contend that we don't really engineer much of anything. Mostly because we don't apply that systems thinking, we are thinking in pockets. We're locally brilliant and globally stupid. We do great in those pockets but we have to do great in how various pockets come together."
Glenn O'Donnell, Forrester
When IT operations has things interrelated, then it's no longer the type of scenario where they have to contemplate how a change will impact a particular system. Relationships are very important in understanding configuration and change management.