powered by CADENAS

Manual

Manual

1.3.9.5. Caching methods for data storage and operation

Caching is basically recommended. In case of doubt make use of CADENAS consulting. Otherwise please regard the default settings.

The following table gives an overview on the different caching methods.

 

FS (Filesystem Caching)

SQUID RFS (Remote Filesystem Caching)

Dependencies

Basic caching for PARTapplicationServer

Addon caching requires PARTapplicationServer

Addon caching requires PARTapplicationServer

Use case

The PARTapplicationServer caches data directly from CADENAS_DATA. Requested data is cached and distributed if requested again.

Also see Section 1.3.9.9.3.7, “ Caching index files of $CADENAS_DATA on PARTapplicationServer ”.

Caching at a remote location.

Remote location caches data which have been requested from the central PARTapplicationServer and distributes it from cache when requested again.

Also see Section 1.3.9.9.2.4, “Caching on client side or at secondary locations ”.

Caching on a client

Client stores all data which have been requested by him and so saves a further request to PARTapplicationServer or CADENAS_DATA.

Also see Section 1.3.9.9.2.4, “Caching on client side or at secondary locations ”.

Cache storage

On the server where PARTapplicationServer is running

On the server at remote location, where SQUID is running

On each client, which has activated RFS caching

Environment

Server installation

Concern solution with several locations

Isolated application for an external employee

Benefit

No direct access on CADENAS_DATA required.

No replication of CADENAS_DATA to remote locations required.

Indexes don't have to be locked for catalog installation. Cache fetches update once performed on CADENAS_DATA.

Performance on data in cache very good.

No replication to the remote locations required.

Performance on data in cache very good, if modelers use the same data.

Performance on data in cache very good.

No direct access on CADENAS_DATA required.

No replication of CADENAS_DATA to remote locations required.

Performance on data in cache very good.