[OSM-talk] advertising, Money: for which purpose and how much?
osm at fop.ns.ca
Wed Jul 4 12:33:31 BST 2007
On Tue, 3 Jul 2007, Martijn van Oosterhout wrote:
> On 7/3/07, Frederik Ramm <frederik at remote.org> wrote:
>> MySQL replication over the net is possible but there will invariably
>> more hiccups than if it's running in the same LAN. Performance of the
>> main server could be adversely affected if there's network congestion,
>> and there's a danger of people becoming hostile to the whole idea of
>> replication because of that.
> What we want is *asynchronous* replication. We don't want the main
> server to ever have to wait for anything else.
MySQL replication is asynchronous; from
"Replication enables data from one MySQL database server (called the
master) to be replicated to one or more MySQL database servers (slaves).
Replication is asynchronous - your replication slaves do not need to be
connected permanently to receive updates from the master, which means that
updates can occur over long-distance connections and even temporary
solutions such as a dial-up service. Depending on the configuration, you
can replicate all databases, selected databases and even selected tables
within a database."
In my experience, replication happens at close to wire speed, and is done
in seconds ("show slave status" gives the number of seconds behind the
master). Last time I set up a replicated MySQL database was for a CRM type
app which stored emails in the database (yes, really); I don't think I
ever saw the slave more than about 5 seconds behind the master. The
databases were connected by a crossover cable talking gig ethernet to each
Synchronous replication would involve using the cluster features, which
require all data to be in memory. I've never had a data set small enough
(or a server that could hold enough RAM) to be able to really test it, and
I don't think it would be applicable here, either.
> As for the turnaround time, most of the serious downloading is
> happening due to renderers and such. They don't care about 2 minutes
> delay. I think even if we keep editors using the main server we'll
> still end up with significant traffic reduction...
I would expect most of the delay to come from internal processing by the
OSM tools. MySQL replication is fast, although it *is*, unfortunately,
prone to going out of sync at the first excuse, and the quickest fix is
generally to scrap the data on the slave and pull it off the master again.
More information about the talk