Coda File System

Re: Large servers ...

From: Peter J. Braam <braam_at_cs.cmu.edu>
Date: Thu, 21 Jan 1999 10:27:08 -0500 (EST)
Wolfram,

Holding databases in a distributed file system requires locking.  Coda has
no facility to do this.  Sorry you'll have to use another system.

For true redundancy you should look into a replicated database, since
there is more to it that merely replicating the log and data file.


- Peter -


On Thu, 21 Jan 1999, Tomalla, Wolfram wrote:

> Hi,
> 
> I'm from a company in germany that produces a controling
> system for a lot of different processes eg. a powerplant
> or a central controling of Coce mashines.
> 
> I thought about running a database (with ODBC and JDBC
> drivers) on a coda filesystem to get a redundant data
> holding. I think the problem will be that there will
> be one or two large files (the database and the logfile)
> and tese files are written to. So if always the whole 
> file is copied to all servers if the database does
> a fsync() it will be nearly impossible to insert any
> data to the database without a timeout. But on the 
> other hand if the database does an fsync() I have to
> be sure, that all changes to the file are copied to 
> the other servers. Else I do not have the redundant 
> data holding I need to keep the date consistend even
> if some hardware crashes.
> I see the only possible solution in adressing and
> copying the file in blocks of eg 4k.
> Even if I look at my private mailfolder I see this 
> problem. I acualy delete my old mail about once a
> month. So there is also the point that closing my 
> mailbox will take several minutes if it has to be copied
> completely over the net to all servers.
> 
> Well I don't have problems in giving the servers for
> our controling system gigabit networking cards, but 
> I don't thing this is the right way to solve this problem.
> 
> Yours sincerly
> 
> Wolfram Tomalla
> 
Received on 1999-01-21 10:27:43