Hi.

I'd like to access (read from and write to) a remote Berkeley DB from
my application. My first thought was to use RPC client/server. But
according to the reference guide, there're significant performance
penalties:

1) "because there is no logical bundling of operations at the server,
performance is usually significantly less than when Berkeley DB is
embedded within the client's address space, even if the RPC is to a
local address." - http://www.sleepycat.com/docs/ref/rpc/client.html

2) "the berkeley_db_svc utility is single-threaded, limiting the number
of requests that it can handle" -
http://www.sleepycat.com/docs/ref/rpc/server.html

Limitation no. 2 is particularly severe, as there won't be *any*
concurrent access.

I'm sure others have been faced with the same problem. What is the best
way to remotely access a Berkeley DB that offers good performance and
scalability? Any suggestions would be greatly appreciated.

On a related note, doesn't DB replication have the same
performance/concurrency problem? In a replicated setup, the Master DB
needs to remotely write to the Client DB, right? Does it use RPC to do
so (and hence face the same performance issue)? Or is there a better
way?

Thanks.

Andy

Well, I'm not really experienced in using Berkeley DB, but I would
guess that a good way is creating a custom protocol, and work over
that. This would give great benefits in some applications, since you
have full control over how and what data is transferred.

Again, I don't really know this, but I think the "significant
performance penalties" aren't that significant. It will probably make
everything much slower, but that's compared to non-remote use. If you
are going to use it over a network, you will get that performance loss
somewhere anyways.

Regards,
/Per Eckerdal