Subject | RE: [ib-support] Varchars and the Network |
---|---|
Author | Leyne, Sean |
Post date | 2001-08-21T22:58:22Z |
Paul,
Rob's presentation is correct.
People often make the mistake of assuming than the storage format is the
same as the transmission format. The reality is that all data fields are
fully 'exploded' by the server and then sent to the client.
AFAIK, the thinking at the time (as far back as v4.0 - 1994?) was based
on the expectation of high-bandwidth/local LAN access, the relatively
low CPU speeds when compared to network transfer and the overall
simplicity of passing the results in the long format.
Of course, progress has continued and these facts have completely
reversed -- low-bandwidth access clients and higher CPU speed.
Also, remember that compression is built-in to a number VPN client
software. So there are existing solutions for the probelm of network
traffic/bandwidth.
There has been some discussion (in the Architecture list) of adding
'over the wire' compression to the client/server library, but nothing
has happened so far.
Sean
Rob's presentation is correct.
People often make the mistake of assuming than the storage format is the
same as the transmission format. The reality is that all data fields are
fully 'exploded' by the server and then sent to the client.
AFAIK, the thinking at the time (as far back as v4.0 - 1994?) was based
on the expectation of high-bandwidth/local LAN access, the relatively
low CPU speeds when compared to network transfer and the overall
simplicity of passing the results in the long format.
Of course, progress has continued and these facts have completely
reversed -- low-bandwidth access clients and higher CPU speed.
Also, remember that compression is built-in to a number VPN client
software. So there are existing solutions for the probelm of network
traffic/bandwidth.
There has been some discussion (in the Architecture list) of adding
'over the wire' compression to the client/server library, but nothing
has happened so far.
Sean