First page Back Continue Last page Summary Graphics
OSS Clustering: An Embarassment of Riches
There are five (and counting) independent general purpose HA systems
There are many independent HP clustering systems
These systems can share little or no code -- developer efforts are diluted, all systems are weakened
There is no particularly good reason for this (except history)
Notes:
I like to use the analogy that people hate the KDE-GNOME split
-- but they're only two of them, and at least they share X windows
In our case, there are 5 mainstream developments, and we share almost nothing!
The History: Many of the systems were originally written as closed source, so they couldn't share any code with other systems. Now that they're open source, they're architected so differently that they have difficulty sharing anything... Basically nothing was designed for reusability...
This greatly dilutes the benefits of the open source methodology -- the "eyes" of the community are spread across far too many lines of code, and the developers re-implement the same thing over and over again.
People who want to use HA software don't know which of the 5 OSS or 20 proprietary packages to be compatable with! Each of these products has their own customer base they won't abandon - for good reason - like they'd go out of business!
This is a messy problem.