I just watched Logica’s presentation on managing large sclae cloud architectures where they shared some lessons learned. All big platforms like LinkedIn, Ebay, Facebook share similar problems when it comes down to performance, scaleability, availability, reliability. So how do you arrive in IT where they are?
1. Change: Adapt to your environment. What works today will not work forever. Recognising that change is needed is a key success factor. Implementing this change even more.
2. Rearchitecture: Optimising will not always be enough. Rearchitecturing needs to be done. And for that you need architects who can do that.
3. One problem at a time: Prioritise problems. Digg deep to find the real reason for the problem and solve that.
4. Solve today’s problem then tomorrows problem: But keep in mind the direction you are going. Don’t go too many steps ahead.
5. Failure is not a option: System failure will happen but plan for it. Create redundancy. Controlling your single failures is controling the failure of the system.
6. Prevention measure: You need to be the first person to discover the problem. If you know that something will fail at a certain point be sure you measure this factor (e.g. if you know that your system can only cope with 500k users then measure the number of users).
7. Look outside the windows: Get inspired from what other people do. Other people might have already solved your problem. Don’t reinvent the wheel.
8. Engage with your audience: Your users and audience will sometimes tell you better what your problem is and they even sometimes solve you problems. Open source is a way of doing so.
9. Team work: Nobody can do it alone.
10. Recruit the right people: It’s better to have 1 good person that 10 wrong people. With the wrong ones it might work, but that’s based on luck.
Anthony Assi’s presentation was really engaging and motivating. Although the single points alone sound simple the true power evolves when they are combined. Great stuff!
From his session at SOA and Cloud Symposium 2010