> jo <jo_yo...[ at ]talktalk.net> wrote innews:6a3f5b6d-bbe6-41f8-b8e0-91bcfd213d51[ at ]a29g2000pra.googlegroups.co
> m:
>
> > I'm pretty new to replication and synchronisation. Having read
> > from a lot of sites about it, I think indirect synchronisation
> > (using TSI Synchroniser) is the method I need to use for the
> > particular situation I'm working with.
>
> > The situation is this. My client has an office with several staff
> > working over a network. The program in question is written in VB6
> > and uses a networked Access 2003 database. A few staff work from
> > home intermittently using their laptops. Currently they have to
> > key in retrospective data when they return to the office. An
> > automated synchronisation of local and remote databases would
> > remove this requirement, particularly as one staff member as a VPN
> > connection from home.
>
> First off, you need to decide if the laptop user needs indirect
> synchronization. I'd say that if the laptop is carried into the
> office and connected to the office LAN, then you don't need any
> indirect synch -- you can simply synch direct (because it's a safe
> connection, i.e., over the LAN).
>
> If you have users editing on their home computers (i.e., desktops,
> not laptops), then you'd need to synch the data in and out, and for
> that you'd definitely need indirect replication over your VPN.
>
> > I feel the first thing I should do is work out how to setup of all
> > the different components used in replication and synchronisation
> > and that is where I have already become unstuck! I wonder if
> > somebody might confirm whether my understanding is correct?
>
> > On the office server I should have a hidden folder for the Design
> > Master mdb,
>
> There's no reason it has to be a hidden folder, but it should be
> somewhere safe. In general, I don't put the DM in a share that users
> have permission to use. A key point: the DM should not be used for
> editing or for regular synching -- it's entire purpose is to be the
> one replica where you can make design changes and you should use it
> *only* for that (and synch it with the other replicas only when a
> design change needs to be propagated, and often enough to keep the
> DM from expiring. Given that the default retention period is 1000
> days, a once-a-month synch should be fine).
>
> > it should have a related shared dropbox folder.
>
> I would say the DM's location is not part of the indirect synch
> topology -- it's only a resource necessary for maintaining the
> replica set, but has no role in regular synchronization.
>
> > Also on
> > the server should be a production replica mdb, shared so as to be
> > visible to the office users. This should also have a related
> > shared dropbox folder.
>
> No machine can have more than one synchronizer or more than one
> dropbox. Since both your DM and your production replica are on the
> same server, you'd have a single dropbox, since you have but the one
> synchronizer. In short, back to the DM question, a DM has no need
> whatsoever for a dropbox, since it's not going to be synched
> indirect (if you want to propagate design changes, you synch with
> the production replica, or with your synchronization hub, and those
> replicas will then propagate the changes to the rest of the replica
> set).
>
> > On a separate networked pc I should have the
> > replication hub mdb in an unshared folder.
>
> Why on a different PC? I don't see why you shouldn't put your
> replication hub on the same server as the production replica.
>
> > It should have a related
> > shared dropbox folder. The remote laptops should be set up each
> > to have a local replica mdb, invisible to the network.
>
> If you're using something other than the server as your replication
> hub, then you need no dropboxes at all on the server, since your
> replication hub (which is on the same LAN, I assume) can synch
> directly with the production replica.
>
> > It should have a
> > related shared dropbox folder that is visible to the network when
> > connected.
>
> Every PC/server on which you have one or more replicas that you want
> to synch indirectly needs to have a synchronizer with its own
> dropbox.
>
> > The server should have Synchroniser continually running, similarly
> > the replication hub pc.
>
> I would disagree strongly. The synchronization hub is on the same
> LAN as the server, so you have no need to use indirect
> synchronization. Thus, you don't need the synchronizer running on
> the server at all.
>
> > Synchroniser should be installed on all networked
> > pcs and the laptops.
>
> Well, only the ones that are going to be carried outside of the
> office and need to be synched indirectly over the VPN while outside
> the office. Any machine that is in the office and connected to the
> LAN (whether all the time or only part of the time) has no need for
> indirect replication and no need for the synchronizer to be
> installed on it.
>
> > Is this a correct understanding of how indirect synchronisation
> > should be setup?
>
> I would say you should have a much simpler replication topology:
>
> 1. keep the synch hub on the server.
>
> 2. don't synch indirect except when outside the office. That will
> tell you which machines need to have the synchronizer installed.
>
> > Is it necessary to have a separate pc for the replication
> > hub?
>
> No. The only reason many people do that is because they don't have
> permission to run the synchronizer on the server 100% of the time
> (which usually requires a locked user logon, something that's now
> very easy with Terminal Server support on all Windows Servers).
>
> > Could the replication hub be on the server?
>
> Absolutely, and that would be preferable, in my opinion.
>
> If you fold the two functions into one server, keep in mind that you
> will manage *only* the synchronization hub. You can set up a regular
> synch schedule between the managed replica(s) and the production
> replica without needing to manage the production replica.
>
> > Any pointers in the right direction would be much appreciated.
>
> Let me know if what I've responded above helps. The key takeaway
> ought to be, I think, that you don't want to get involved with the
> administrative overhead of implementing indirect replication (and
> installing the synchronizer and making sure it's running and all
> that) except for the PCs that are going to be synched with the
> mother ship over the VPN from outside the office. Any machine that
> connects to the office LAN can use the much simpler direct
> replication.
>
> --
> David W. Fenton
http://www.dfenton.com/> usenet at dfenton dot com
http://www.dfenton.com/DFA/