A 2009 Forrester study found that 47% of users expect a web page to load in two seconds or less and 40% will abandon a web page after three seconds if it still hasn’t loaded. This doesn’t leave businesses with much margin for error. End users seeking to access information via the web are not concerned with how this data is delivered to them, they just want and expect a fast response and if they don’t get it they are prepared to go to go elsewhere.
The mainframe isn’t dead
New applications such as e-commerce and mobile apps are likely to be hosted on modern technology platforms such as the cloud, but they still need to draw information from the mainframe. For example, a mobile banking application will still need to connect with the CRM system in order to supply a customer’s bank balance. With increased usage and a rising focus on information access at any time, from any place, both the CIO and the mainframe are under pressure to make sure that applications meet business and customer expectations.
In order to meet these end user demands, changes are constantly being made to applications. When making these changes IT can unwittingly impact performance by introducing amends that create faults. These might not stop the application working, but they could make it inefficient. For example, rather than the application seeking to pull information from the mainframe once, it does so 50 times.
Due to the complex nature of IT landscapes it is not unlikely that this fault in the code will go unnoticed. These inefficiencies erode application performance and increase the strain being put on the mainframe, causing excessive CPU consumption and greater hardware and software costs.
MIPS management maturity
Looking at how the performance of applications that rely on the mainframe can be improved not only reduces CPU consumption and response times, but also lowers the lifetime costs of applications, underpins a better customer experience and ultimately drives revenue. Typically mainframe application performance primarily focuses on optimising MIPS consumption. When MIPS usage rises, excessive CPU consumption forces expensive hardware upgrades.
There are many different factors that dictate how far down the maturity scale companies are when it comes to MIPS management. But just like a martial arts student only progresses to becoming a black belt by passing through various levels of techniques and processes, the same is true for MIPS management. We believe there are five levels of MIPS Maturity:
- Level one – Reactive (chaos)
- Level two – Repeatable (proactive)
- Level three – Defined (process oriented)
- Level four – Managed (disciplined)
- Level five – Optimising (Kaizen)
By default most companies are at level one. Level one is classified as reactive or chaos, simply because the majority of time is spent firefighting; whereas at level two companies have some level of repeatable methods for reducing excessive resource demands on the system.
Very few organisations find themselves in a position where MIPS management is embedded within the organisation from the top down (level five). Companies struggle to get to this stage because there is very little guidance out there to help them end the self-perpetuating cycle of MIPS overspend. What is needed is a step-by-step guide to help enterprises navigate the best way for their organisation to maximise performance and identify waste.
The road to MIPS Management
In order to help their business progress through the MIPS Maturity Model in the most effective way, IT needs to contemplate the following factors:
- Consider the objectives: Any project aimed at reducing MIPS usage will not perform as hoped unless quantifiable, specific and well-documented objectives are put in place. Goals might include reducing CPU cycles consumed by an application or decreasing the frequency of application failure
- Define the project scope: At first a MIPS management project can appear very daunting. It is therefore key to focus the project on what matters most and where the most gains can be made. It makes sense to identify those applications that are consuming the most CPU and the associated tasks that fail most often because in doing so they consume the most resources. This information will help to sharpen your focus in relation to the objectives you set out at the beginning
- Collect the data: Once the above step has been completed it should be possible to create a ‘hit list’ of failing tasks that consumed the most CPU seconds or had the longest time elapse, as these are draining the most resources. Ideally you need to create reusable groups of measurement requests, which will allow you to identify trends based on certain environments, peak times etc
- Analyse the data: By looking at the data gathered it will become clearer what steps need to be taken in order to reduce the volume of resources being consumed by applications
- Make changes: To improve performance and reduce faults, it is necessary to make changes that reduce the demands an application places on the system and to prevent faults from occurring. Resolving failures in high consuming and failing tasks should be the priority, as this results in two immediate benefits. Firstly, the application no longer fails, avoiding costs associated with recovery and rerun. Secondly, CPU consumption is reduced due to a decrease in inefficiencies
- Reassess: To determine what impact the changes have made on the application’s performance, IT needs to look at the measurement requests it originally set up in order to compare and contrast
- Report and reward: All too often when projects are started, little thought is given to how success is reported back to the business and similarly, how those involved should be rewarded to avoid poor morale. Avoid this by putting KPIs in place for team members so they know what they are striving towards
- Establish continual MIPS management processes: When initial goals have been reached, the next challenge is to establish the standards and processes needed to integrate MIPS management into the application lifecycle, to ensure that application resources are used effectively in the future
At a time when most companies are looking to eradicate unnecessary spend, mainframe services offer a significant window of opportunity. Yet what has been missing to date is a dedicated ‘plan of action’ that practically helps companies to manage MIPS usage. Certainly there isn’t a silver bullet, but effective MIPS Management can make a big difference to the competitiveness of a business as well as its bottom line.