| | 1 | [[PageOutline]] |
| | 2 | = Cache Management = |
| | 3 | |
| | 4 | Because network connections are expensive in time, compute cycles and network bandwidth, a strong caching paradigm (love using that word) was implemented to help lower the use of resources and lower the response time. |
| | 5 | |
| | 6 | == Configuration == |
| | 7 | - the timeout and size can be tweaked as needed to support a particular environment. The system will send out an INFO alert if it's purging too quickly. |
| | 8 | - the memory cache can be completely disabled by setting the size to 0. This should only be done if you are using a single process, as the database interface *should* be caching results. |
| | 9 | == Lookups == |
| | 10 | - Each search to the AD server is through a cached LDAP filehandle. The restart option has been enabled to allow it to persist. |
| | 11 | - When valid data is found .. the data is pickled, and the key hashed and stored in a new table in the database called ad_cache. |
| | 12 | - It's also stored in a class variable {{{_cache}}} which is a dictionary. |
| | 13 | |
| | 14 | == Retrieval == |
| | 15 | - the memory cache is checked first |
| | 16 | - then the database cache |
| | 17 | - lastly the AD server itself is searched |
| | 18 | - Any results are then pushed back on the cache(s) |
| | 19 | |
| | 20 | == Integration == |
| | 21 | - data is cached by 'class' where appropriate to speed up processing, |
| | 22 | - all ldap searches are cached using a key which is a hashed combination of ( base_dn, scope, filter ) |
