Sarquol Ltd

I have provided an updated performance model on my web site. This includes a number of improvements, mainly centred on data growth modelling and the analysis of function usage where different functions have significantly different usage times within a day. It is available under the “Free Resources” section of the web site.

The process of assuring the performance of systems can be expensive. The costs involved can include software development, implementation and maintenance of monitoring systems and hardware investment. It can be easy to forget, however, that a system that performs well has higher value to the business than one that performs badly. Thus, if there is no critical performance problem that needs resolution it can be seen as an ineffective use of resources. Read the rest of this entry »

A performance model will be full of assumptions and estimates, and yet it is necessary to make important design decisions and capacity choices based on its predictions. This means that it will be necessary to check that it is correct, known as validating the model. Read the rest of this entry »

Again, this has come from the internet so I can’t provide its origins. Out of interest, my wife sent it to me:

Married 40 some odd years, I took a look at my wife one day and said, “Honey, years ago, we had a cheap apartment, a cheap car, slept on a sofa bed and watched a 10 inch black and white TV, but I got to sleep every night with a hot 25 Year old blond. Now, we have a nice house, nice car, big bed and plasma screen TV, but I’m sleeping with a 60 year old woman. It seems to me that you are not holding up your side of things. Read the rest of this entry »

It is interesting to speculate on the future once in a while. It isn’t so long ago that Intel was talking about future chips running hotter than the surface of the sun. Now IBM is trying to run chips at near absolute zero:

500GHz frozen chips on the menu at IBM – vnunet.com

It is looking like the data centre of the future is going to be a very inhospitable place. Maybe it will be practical to place them in a satellite and place them in orbit. The cost would be significant, but there are advantages too. In particular there is a lot of solar power available up there. There are a few problems with the plan though – network latency will be a significant issue. Disaster recovery planning will take on a few interesting aspects as well. OK, I’ll admit it this one is probably a few years away from delivery as a practical business plan. If you have some more creative ideas for the location of future data centres e-mail me or add them as comments when this bulletin appears on the web site. Let me know if I can use them in future editions, but also let me know if you want me to keep your name off it.

In the bulletin so far I have mainly concentrated on the subject of Capacity Management as a whole. On engagements with clients, however, it is quite common for me to be called in once there is already an identified performance problem. The good news is that the best approach to resolving a performance problem that already exists is very similar to the Capacity Management already discussed elsewhere. The bad news, however, is that this means going through similar processes in a much reduced timescale. The chances of managing this are greatly improved if the original system project used a capacity management approach on implementation. Read the rest of this entry »

Let us assume that you have started a piece of performance modelling. You have chosen a set of key transactions, or better still have been told what is most important by the business. You may have an idea of the performance levels that the business need (more on this another time), and you start working with the business to work out the volumes that they are expecting to put through the system. To do the appropriate performance modelling you will need to define the behaviour of a “typical user” and then scale this up to produce a total behaviour profile for all users. It is unfortunate that you are unlikely to receive figures that can be used directly from your business users. You have the following choices at this point: a) educate them in the modelling you are doing and the format you need the numbers in, or b) take the numbers they have provided and reproduce them from your modelling. Read the rest of this entry »

Since we are coming into the holiday period I will divert on to a little humour. This is an extract of a list doing the rounds on the internet, and so I can’t provide its source. E-mail me if you want the full list:

1. My mother taught me TO APPRECIATE A JOB WELL DONE.

“If you’re going to kill each other, do it outside. I just finished cleaning.”

Read the rest of this entry »

If you do not already have a performance testing tool and are doing HTTP or HTTPS development then I would recommend that you look at OpenSTA. The tool is open source, and so the only up front cost involved is the time it takes to learn the interface and produce scripts. It isn’t the only open source load testing tool out there, but it has a strong combination of simplicity of use and scripting flexibility. Read the rest of this entry »

A new computer system goes into production and starts with a pilot. A set of load testing is done to make sure that the live user levels can be supported, with some problems. These problems are resolved and maybe some extra hardware purchased to ensure performance at live volumes is acceptable, and then a full roll out is started. The project team are on site for the first few weeks, and resolve the problems that are experienced initially. The system is handed over to support along with a development team member and then the project team is disbanded. The support team are left to complete the roll out process, and the development team member moves on after a while. Read the rest of this entry »