Caching of EJBHome and EJBObject

Question

Hi,
Currently I am using Weblogic 5.1 and running in the NT environment.
I have a Session Bean (A) which needs to access another Entity or
Session Bean, the usual practice is to create a JNDI context and do a
JNDI lookup in the Session Bean (A) to get the EJBHome reference of the
other Entity or Session Bean.
1. To improve the performance, we would like to do a one time lookup to
get the EJBHome reference and cache the EJBHome reference inside the
Session Bean (A) in private variables which would be initialized in the
constructor.
Will doing this cause any problem when accessing the cahched EJBHome in
a non-clustering and clustering environment ?
2. To improve the performance, we would like to cache the reference to
the EJBObject of the other Entity Bean or Session Bean inside the
Session Bean (A) in private variables which would be initialized in the
constructor.
Will doing this cause any problem when accessing the cahched EJBHome in
a non-clustering and clustering environment ?
Thanks.
Regards
Terry

Answer 1

Terry,
Hope this helps.
Kev.
Terry wrote:
Hi,
Currently I am using Weblogic 5.1 and running in the NT environment.
I have a Session Bean (A) which needs to access another Entity or
Session Bean, the usual practice is to create a JNDI context and do a
JNDI lookup in the Session Bean (A) to get the EJBHome reference of the
other Entity or Session Bean.
1. To improve the performance, we would like to do a one time lookup to
get the EJBHome reference and cache the EJBHome reference inside the
Session Bean (A) in private variables which would be initialized in the
constructor.
Will doing this cause any problem when accessing the cahched EJBHome in
a non-clustering and clustering environment ?All bean homes are clusterable. This means if you are issuing a method call
off of home and home experiences problems (network or otherwise), an
exception will be generated and caught by a home replica aware stub. Once
caught the method call is reissued on replicated home within another server
instance. (Being clustered/non-clustered is transparent except to the extent
in clustered there is the concept of a stub replicate refresh interval which
defautlts and can be changed. This simple controls how often you replica
aware stub is updated. Ditto for 2 below. )
>
>
2. To improve the performance, we would like to cache the reference to
the EJBObject of the other Entity Bean or Session Bean inside the
Session Bean (A) in private variables which would be initialized in the
constructor.
Will doing this cause any problem when accessing the cahched EJBHome in
a non-clustering and clustering environment ?I don't see any problems. Be aware not all beans can be faild over or are
load levelable. You need to see:
http://www.weblogic.com/docs51/cluster/concepts.html#1024225 for a
definitive explanation of this. Statefull session beans and read/write
entity beans introduce special problems and consequently require client
control of the process.
>
>
Thanks.
Regards
Terry