Providing Affordable IT Management to SME's Application sizing and modelling
Although two of the lesser used sub processes of capacity management, these offer great opportunities to prevent system down time as they are forward looking and prepare you for what might happen. The easiest way to distinguish between them is to thing of Application Sizing as something you do BEFORE implementing a new system whilst Capacity Modelling is something done ONCE the system is in and established.
So lets look as Application Sizing first…..
You decided to buy a new photo editing and storage program. What things do you need to consider? Well firstly you will need to look at the basic system requirements. This may consider things like operating system level but from a capacity viewpoint will also consider requirements including processor speed, memory requirements and Hard Drive space. Once we have looked at the basic‘s we could move onto to data transfer. The program may store files in different formats (eg accepting RAW images). This may increase the size of the files and if transferring them via the internet, you may find that your service does not have the required bandwith or your email system may have limits on the size of the file. Finally, the introduction of new functionality or file types may increase your storage needs. At this point you may need to review your current available space and consider options to increase this (such as external hard drives). This example illustrates a typical walk through of application sizing and whilst various combinations of application and infrastructure types exist, the principle remains the same. The key aim is to consider all aspects of the introduction of a new application to ensure that capacity issues are addressed prior to implementation.
Capacity Modelling is a process that is based around “what if” scenarios. A real life example from a high street retailer is illustrated as follows: Following the introduction of Chip And Pin credit cards, the data file sent to the banks now contains additional information. This resulted in the file being 3 times larger. This was not a problem as the system had a 3hr dialing window to contact the bank and transmit the data. At the end of the window the connection terminated whether the data had completed or not. On the final week of Christmas 2005, the dial did not completed and the retailer lost several days interest. Why was this ? Quite simply in the final week of Christmas sales went up which meant more people used credit cards. This meant that the data file was bigger. In previous years their had been no problem getting this information to the bank in the 3 hr window but as Chip and Pin had trebled the size of the file, it failed.
So how does this fit in with Capacity Modelling? Well going into Christmas the variable’s were known as follows:
1) The size of a data file for one transaction
2) The budget for Christmas week
3) The percentage of people that would probably shop via credit card
4) The normal transmitting times / rates
This would give the Capacity Manager enough data to estimate the size of the data file and also to model that nights transmission to see if it would fit. That is the basic principle of Capacity Modelling – taking known data, applying it to a set of given circumstances to see if a service breach would occur. Hopefully the overview of these two processes gives SME’s a few extra tools to use during the day to day management of current systems or things to consider during application upgrades.
3rd Aug 2009