In another thread, I began to address a performance issue with PostgreSQL. Idea came to me, why not add support for zabbix in a key-value database?
I know it's a challenge for a lot of work and the remaking of the concept of storing data. But I think more power and opportunity to better scaling arguments are good for.
Relational databases for information data with complex linkages between them is great. For the measured data I do not think so. What we have here for information? Time series data. In most other monitoring systems, the only available "database" for this type of data - RRD.
I found an example for which similar data is also used as key-value databases, namely: www.serverdensity.com
I do not have more experience with key-value Databases, but I think it might be an interesting idea.
What do you think?
I know it's a challenge for a lot of work and the remaking of the concept of storing data. But I think more power and opportunity to better scaling arguments are good for.
Relational databases for information data with complex linkages between them is great. For the measured data I do not think so. What we have here for information? Time series data. In most other monitoring systems, the only available "database" for this type of data - RRD.
I found an example for which similar data is also used as key-value databases, namely: www.serverdensity.com
I do not have more experience with key-value Databases, but I think it might be an interesting idea.
What do you think?
Comment