| 摘要 |
The SNMP cache of the present solution supports multi-core/multi-node environment by recalculating the SNMP ordering of the entities in the response from multiple cores/nodes at insertion time. The most significant gain is achieved by prefetching or augmenting the cache, wherein while requesting an entity and its stat information, next few entities in SNMP order are requested from the owner processes. SNMP Management systems extensively utilize repeated GETNEXT (such as via a SNMP WALK) and few next responses may be served from the cache directly. Further performance improvements are obtained by introducing another level of cache on top of the existing cache. This auxiliary cache ensures a high hit ratio for repeated SNMP GETNEXT request (SNMP WALK operation) by caching last accessed entity within the main cache. This auxiliary cache also aids in insertion in the larger main cache by maintaining pointers to last accessed entity before the main cache miss. Cache implements other features like new stat inclusion/updating of the already cached entity. |
