On 8/24/2011 9:17 AM, Jaap Winius wrote: > But, wouldn't that cause trouble in the mornings when those workstations > are turned on again and a number of them try to re-synchronize their 1GB > caches across my relatively puny 6Mbps WAN links?
Caches are not actively sync'd. Caches are passive data stores that are used to avoid copying data across the network when the data is known to already exist locally. If the required data is not present in the cache, then it is read from the file server. If the cache is too small, more data is read from the file server than would otherwise be necessary. Cache coherency is maintained in AFS via the use of status information which has an expiration time and a callback protocol that is used by the file server to notify the client when the unexpired status information has been altered. AFS does not read and cache whole files. It reads file chunks which are tracked by File ID, File Offset, and Data Version. Data is used from the cache when the File ID, File Offset and Data Version required to satisfy an application request match. Otherwise, the data is read from the file server. For read/write data, if the cache is too small, the cache manager is required to flush data to the file server sooner than it would prefer. Since many files used today are in the GB range, it is not unusual to have caches sizes of 10GB to 20GB. The local disk is cheap; network bandwidth is not. Jeffrey Altman
signature.asc
Description: OpenPGP digital signature
