In instances where you have a high-latency in the network, running
synchronous block-level replication will simply give you abysmal
performance. You can't get something for nothing.
Block-level replication over WANs are usually acomplished asyncronously. In
most iSCSI implementations I've seen (such as Sanrad) you take a snapshot,
replicate the snapshot, then throw the snapshot away.... wait, rinse,
repeat.
In MSSQL/Exchange scenarios, it would be much more reasonable to simply use
application layer replication as it is specifically designed to address
these issues.
-=dave
----- Original Message -----
From: "Robert Neuschul" <[EMAIL PROTECTED]>
To: <[email protected]>
Sent: Friday, May 05, 2006 9:26 AM
Subject: Re: [OF-users] Openfiler as a SAN
Rafiu
Absolutely.
Sure thing. We're going to be adding replication management to the mix
in 3Q06.
That's very good news: what I'm wondering is how resilient the whole
architecture might be when deploying replica SQL or Exchange stores over a
publicly routed network. Thus for example if one wanted to run real-time
replication of a SQL or Exchange server to another site on the other side
of the country using SDSL or ADSL then how well will the MS iSCSI
initiator
and OF itself jointly cope with potentially very variable latencies etc.
I suppose it would be premature to speculate about this <hopefully> but do
you have any ideas? </hopefully>
TIA
Robert.
_______________________________________________
Openfiler-users mailing list
[email protected]
https://lists.openfiler.com/mailman/listinfo/openfiler-users
_______________________________________________
Openfiler-users mailing list
[email protected]
https://lists.openfiler.com/mailman/listinfo/openfiler-users
