On 20/11/2007, Yonah Russ <[EMAIL PROTECTED]> wrote: > On Nov 20, 2007 7:32 AM, Amos Shapira <[EMAIL PROTECTED]> wrote: > > > > > On 20/11/2007, Yonah Russ <[EMAIL PROTECTED]> wrote: > > > Hi, > > > > > > AFAIK MySQL cluster requires using the NDB engine which is not 100% > > > compatible with MYISAM, etc. The more recent a version you have the > better > > > off you will be but there was one point where autoincrement wasn't even > > > supported. Also from what I know, a NDB cluster needs at least three > nodes. > > > > Excuse me for my ignorance but why should I care about MyISAM > > compatibility? Can't I use InnoDB as I currently use (or actually was > > dictated by the guys who did the initial web application for us)? > > > > > > No- as I said you need to use the NDB engine which will have compatibility > issues with InnoDB as well. MyISAM was just an example. > > > > > > > > > > > That doesn't mean you can't have an active-active database. If you > > > application is ready for it, you can set up MySQL in a Master Master > > > configuration (aka multimaster). For your application to be compatible > you > > > basically need to use only autoincrement primary keys for all your > tables > > > but you should read the whole Megilah - here is a good article: > > > > http://www.onlamp.com/pub/a/onlamp/2006/04/20/advanced-mysql-replication.html > > > > Thanks for that pointer - after half a day of googl'ing around about > > MySQL cluster, the link above was the first one to point to the fact > > that MySQL cluster is an in-memory database. > > > > > > > > > > Google has also released some code for multimaster configurations and > they > > > have some docs also: > > > http://code.google.com/p/mysql-master-master/ > > > > Thanks. I read about such a method (with 4 nodes) almost a year ago > > and thought that it might have become obsolate with MySQL cluster. Now > > I'll have to re-consider it. > > > > > > > Regarding Postgres- I can't recommend anything specific but I am looking > > > into pgpool and pgpool II for a new project. > > > > Why is pgpool relevant here? It more kind of a "reverse proxy fo > > postgres" in that it'll hide server failure by auto-connecting to > > another server. Does it handle data synchronisation as well? > > > > From the pgpool II web page: > pgpool-II is a inherited project of pgpool (to classify from pgpool-II, it > is sometimes called as pgpool-I). For those of you not familiar with > pgpool-I, it is a multi-functional middle ware for PostgreSQL that features > connection pooling, replication and load balancing functions. > > As I said, I'm only evaluating it now but if it does what it says, then it > might be useful.
Ah I see now and it brought back what I learned about it a while ago - what it does is to rewrite every SQL statement which passes through and modifies the database and sends it to all the servers in parallel. Read-only statements are passed to one server only. Pretty nifty. Back when I learned about this pgpool(-1?) was still finding and cleaning out bugs in correct SQL code replication, and I'm not sure how much in sync where the servers in the boundary cases. > > > > Another point which was just raised in the office - it appears that we > > are going to use DRBD for SQLite synchronization - if so then it > > should make sense to use DRBD for MySQL synchronization as well to > > reduce the configuration complexity. Any thoughts on that? > > > > I have only used DRBD in Active-Passive settings to eliminate the need for > replication. > Maybe you could put the database files on a cluster filesystem over DRBD? That's what I saw to be suggested in a few places, can't remember which of the databases was mentioned with it. I saw this (cluster-aware filesystem over DRBD) also mentioned in the context of sharing Xen images for fail-over (so you have a Xen filesystem (over LVM?) over DRBD and if the Xen host crashes then another host can pick up its guests and run them from the same LVM). My take-away form this is that DRBD+Cluster-FS might be used as a cheap replacement for a real SAN. > > Depending on your budget/configuration, NetApp has recently started selling > an inexpensive storage solution ($3000) the StoreVault S300. I haven't tried > it but the price is right. > I've also been meaning to try Coraid storage although their prices in Israel > were not as attractive as in the States. Thanks for the heads-up. Our host is in the US so it might be relevant. Cheers, --Amos ================================================================= To unsubscribe, send mail to [EMAIL PROTECTED] with the word "unsubscribe" in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
