Skip site navigation (1) Skip section navigation (2)

Re: Clustering & Load Balancing & Replication

From: "Shoaib Mir" <shoaibmir(at)gmail(dot)com>
To: "Andy Dale" <andy(dot)dale(at)gmail(dot)com>
Cc: "org(at)kewlstuff(dot)co(dot)za" <org(at)kewlstuff(dot)co(dot)za>, pgsql-general(at)postgresql(dot)org
Subject: Re: Clustering & Load Balancing & Replication
Date: 2006-12-26 15:21:22
Message-ID: bf54be870612260721n5c156104tffa3f3d927a061d6@mail.gmail.com (view raw or flat)
Thread:
Lists: pgsql-general
>>We are trying to achieve High Availability over load balancing, so
basically we always try and have 2 databases in the same state while >>both
are active,

What problems do you see with Slony + Linux HA combo there? I think a Slony
failover can do the same and promote a slave node to master in case of a
master node failure, and then you can actually do the load balancing by
using the slave nodes in the system.

Please let me know on any updates for OpenSSI if you get a chance to try
that out

-------------------
 Shoaib Mir
EnterpriseDB ( www.enterprisedb.com)

On 12/26/06, Andy Dale <andy(dot)dale(at)gmail(dot)com> wrote:
>
> We are trying to achieve High Availability over load balancing, so
> basically we always try and have 2 databases in the same state while both
> are active, and if one goes down it should (hopefully) failover seemlessly.
> Thanks for the info on OpenSSI, as i had not heard of this before and will
> look into it further.
>
> Cheers,
>
> Andy
>
> On 26/12/06, Shoaib Mir <shoaibmir(at)gmail(dot)com> wrote:
> >
> > Why are you going for a multimaster case here? are you doing it for load
> > balancing? if then you can also do it horizontally with a multi disk
> > setup... Slony can be a real good candidate here as well with Linux HA in
> > combination.
> >
> > For going on a vertical solution you can try OpenSSI and see if that can
> > work for you, haven't tried that myself but will like to hear about
> > PostgreSQL configuration with OpenSSI
> >
> > -------------------
> > Shoaib Mir
> > EnterpriseDB ( www.enterprisedb.com)
> >
> > On 12/26/06, Andy Dale < andy(dot)dale(at)gmail(dot)com> wrote:
> > >
> > > The company i am working for has a trail/evaluation license for
> > > p/cluster, but unfortunately i cannot get it to function as a JBoss
> > > datasource.
> > >
> > > Cheers,
> > >
> > > Andy
> > >
> > > On 26/12/06, Shoaib Mir <shoaibmir(at)gmail(dot)com> wrote:
> > > >
> > > > Yes, that is true with pgpool. I did face the same as well.
> > > >
> > > > There is another as well Uni-Cluster (http://www.continuent.com/index.php?option=com_content&task=view&id=213&Itemid=170
> > > > ), haven't tried yet but it might help you there...
> > > >
> > > > ---------------
> > > > Shoaib Mir
> > > > EnterpriseDB ( www.enterprisedb.com)
> > > >
> > > > On 12/26/06, Andy Dale <andy(dot)dale(at)gmail(dot)com> wrote:
> > > > >
> > > > > The issue i had with pgpool (1 or 2) was that (correct me if i am
> > > > > wrong) you had to start the pgpool cluster with both nodes in the same
> > > > > state.  I thought this would mean that if you had a DB fail, before you
> > > > > could re-introduce it into the pgpool cluster you would have to manually
> > > > > sync it with the cluster state, is this correct ??
> > > > >
> > > > > The system i need multi master sync for is highly transactional,
> > > > > so if the behaviour i stated above is correct it is not suitable.  I have
> > > > > tried s-lony, and while i was pleased with the performance, it is only
> > > > > Single Master - Multi Slave which is not acceptable as well.
> > > > >
> > > > > Andy
> > > > >
> > > > > On 26/12/06, Shoaib Mir <shoaibmir(at)gmail(dot)com > wrote:
> > > > > >
> > > > > > pgpool-II might help you there too I guess...
> > > > > >
> > > > > > ---------------
> > > > > > Shoaib Mir
> > > > > > EnterpriseDB ( www.enterprisedb.com)
> > > > > >
> > > > > > On 12/26/06, Andy Dale <andy(dot)dale(at)gmail(dot)com> wrote:
> > > > > > >
> > > > > > > Hi,
> > > > > > >
> > > > > > > I have just read the statement that Postgres does have (with
> > > > > > > end user assembly) multi-master replication system.  Is this just PGCluster
> > > > > > > or something else ? if it is not PGCluster, then how can this be achieved ?
> > > > > > >
> > > > > > > Cheers,
> > > > > > >
> > > > > > > Andy
> > > > > > >
> > > > > > > On 24/12/06, Shoaib Mir < shoaibmir(at)gmail(dot)com> wrote:
> > > > > > > >
> > > > > > > > I guess the latest 8.2 Windows PostgreSQL installer does
> > > > > > > > come with a Slony option and you can set it up easily using pgadmin too.
> > > > > > > >
> > > > > > > > This link -->
> > > > > > > > http://people.planetpostgresql.org/xzilla/index.php?/archives/200-Alpha-testing-Slony-on-win32-Crib-Notes.htmlmight help you as well.
> > > > > > > >
> > > > > > > > -----------------
> > > > > > > > Shoaib Mir
> > > > > > > > EnterpriseDB (www.enterprisedb.com)
> > > > > > > >
> > > > > > > > On 12/25/06, org(at)kewlstuff(dot)co(dot)za <org(at)kewlstuff(dot)co(dot)za>
> > > > > > > > wrote:
> > > > > > > > >
> > > > > > > > > Thanks Chris,
> > > > > > > > > I see you a core member of Slony team and a replication
> > > > > > > > > guru so I'll look
> > > > > > > > > into it.
> > > > > > > > > I'm not slamming Slony I think its probably the right tool
> > > > > > > > > for type of work
> > > > > > > > > your company Afilias does. Just wish you would make an
> > > > > > > > > official Windows
> > > > > > > > > version of Slony as well.
> > > > > > > > > Anyway thanks for the education, and I think it would be a
> > > > > > > > > good thing if
> > > > > > > > > your site on replication, was also listed on Postgresql...
> > > > > > > > > good research.
> > > > > > > > > Merry Xmas
> > > > > > > > >
> > > > > > > > > ----- Original Message -----
> > > > > > > > > From: "Christopher Browne" <cbbrowne(at)acm(dot)org>
> > > > > > > > > To: < pgsql-general(at)postgresql(dot)org>
> > > > > > > > > Sent: Sunday, December 24, 2006 4:23 AM
> > > > > > > > > Subject: Re: [GENERAL] Clustering & Load Balancing &
> > > > > > > > > Replication
> > > > > > > > >
> > > > > > > > >
> > > > > > > > > > Centuries ago, Nostradamus foresaw when
> > > > > > > > > org(at)kewlstuff(dot)co(dot)za would write:
> > > > > > > > > >> Suggest you download my little application and read the
> > > > > > > > > documentation,
> > > > > > > > > >> you'll see its very different, maybe even interesting.
> > > > > > > > > >> Maybe they should change that to.... Postgres DOES HAVE
> > > > > > > > > a free
> > > > > > > > > >> multi-master
> > > > > > > > > >> replication system :)
> > > > > > > > > >
> > > > > > > > > > It isn't systematically usable as such, without a whole
> > > > > > > > > lot of
> > > > > > > > > > end-user assembly.
> > > > > > > > > >
> > > > > > > > > >> One comment they make.... "Heavy write activity can
> > > > > > > > > cause excessive
> > > > > > > > > >> locking,
> > > > > > > > > >> leading to poor performance. In fact, write performance
> > > > > > > > > is often worse
> > > > > > > > > >> than
> > > > > > > > > >> that of a single server. Read requests can be sent to
> > > > > > > > > any server."
> > > > > > > > > >> I'm not sure I agree with that... or maybe MVCC is just
> > > > > > > > > fantastic.... I
> > > > > > > > > >> tested it.
> > > > > > > > > >> The 2 phase commit locking is definitely happening at
> > > > > > > > > record level, so
> > > > > > > > > >> only
> > > > > > > > > >> if the multimasters all hit the same record is there
> > > > > > > > > the potential for
> > > > > > > > > >> lock
> > > > > > > > > >> conflict.
> > > > > > > > > >> Why will dB's being randomly used, hit the same
> > > > > > > > > records, I think its a
> > > > > > > > > >> low
> > > > > > > > > >> probability to begin with?
> > > > > > > > > >
> > > > > > > > > > That's only true if you are certain that the update
> > > > > > > > > pattern is NOT
> > > > > > > > > > involving a shared set of records.  IN GENERAL, heavy
> > > > > > > > > write activity
> > > > > > > > > > can cause locking to become mighty expensive, which is
> > > > > > > > > certainly a
> > > > > > > > > > true statement.
> > > > > > > > > >
> > > > > > > > > >> Not happy with that, I wrote a multithreaded routine
> > > > > > > > > and got them to all
> > > > > > > > > >> smack the same record, it NEVER ROLLED BACK, and if
> > > > > > > > > there is performance
> > > > > > > > > >> degradation, I didnt notice it... again probably a
> > > > > > > > > testament to the MVCC
> > > > > > > > > >> design.
> > > > > > > > > >
> > > > > > > > > > It seems likely to me that this requires some careful
> > > > > > > > > validation of
> > > > > > > > > > testing.
> > > > > > > > > >
> > > > > > > > > > An effect we see is that if a set of transactions are
> > > > > > > > > "fighting" over
> > > > > > > > > > a single "balance" record, they will essentially
> > > > > > > > > serialize over that.
> > > > > > > > > >
> > > > > > > > > > On a system with a single CPU, it is not obvious that
> > > > > > > > > you'll see a
> > > > > > > > > > degradation there because, since you only have the
> > > > > > > > > single CPU, it
> > > > > > > > > > would be serializing the activity anyways.
> > > > > > > > > >
> > > > > > > > > > Try it out on an 8-way SMP system and you may see things
> > > > > > > > > differently.
> > > > > > > > > >
> > > > > > > > > >> In any event if you look at the documentation, you'll
> > > > > > > > > see SPAR is not
> > > > > > > > > >> multimaster or nothing. Can use say one server in an
> > > > > > > > > office and another
> > > > > > > > > >> to
> > > > > > > > > >> pump data to a remote web site... not sure if you would
> > > > > > > > > even call that
> > > > > > > > > >> multimaster, thats the point, I'm not sure SPAR fits
> > > > > > > > > any pure theory
> > > > > > > > > >> category.
> > > > > > > > > >
> > > > > > > > > > There are a few tests I could throw at it that tend to
> > > > > > > > > challenge
> > > > > > > > > > replication systems vis-a-vis "fidelity of results."  I
> > > > > > > > > otta see if I
> > > > > > > > > > can find them in a readily deployable form.
> > > > > > > > > >
> > > > > > > > > > There are two notable anomalies which have been known to
> > > > > > > > > break
> > > > > > > > > > replication systems:
> > > > > > > > > >
> > > > > > > > > > 1.  Nondeterministic updates:
> > > > > > > > > >
> > > > > > > > > > For instance, functions that are nondeterministic:
> > > > > > > > > >
> > > > > > > > > >  insert into rtable values (random(), now());
> > > > > > > > > >
> > > > > > > > > > Or result sets that are nondeterministic:
> > > > > > > > > >
> > > > > > > > > >  insert into rtable2 (select * from mytable where
> > > > > > > > > some_attr='foo'
> > > > > > > > > >     order by random() limit 5);   -- Where there are 25
> > > > > > > > > records with
> > > > > > > > > > some_attr='foo'
> > > > > > > > > >
> > > > > > > > > > 2.  Value swapping:
> > > > > > > > > >
> > > > > > > > > > Consider the table:
> > > > > > > > > >
> > > > > > > > > > create table t1 (mk integer primary key, val text unique
> > > > > > > > > not null);
> > > > > > > > > >
> > > > > > > > > > insert into t1 (mk, val) values (1, 'chris');
> > > > > > > > > > insert into t1 (mk, val) values (2, 'dave');
> > > > > > > > > > insert into t1 (mk, val) values (3, 'brad');
> > > > > > > > > >
> > > > > > > > > > begin;
> > > > > > > > > > update t1 set mk = 99 where mk = 1;
> > > > > > > > > > update t1 set mk = 1 where mk = 3;
> > > > > > > > > > update t1 set mk = 3 where mk = 99;
> > > > > > > > > > commit;
> > > > > > > > > >
> > > > > > > > > > Is there a condition where a pause somewhere in there
> > > > > > > > > will cause
> > > > > > > > > > replication to break?  Note that there have been
> > > > > > > > > replication systems
> > > > > > > > > > (erServer) that this set of updates can, intermittently,
> > > > > > > > > cause to fall
> > > > > > > > > > over.
> > > > > > > > > > --
> > > > > > > > > > let name="cbbrowne" and tld=" linuxfinances.info" in
> > > > > > > > > String.concat "@"
> > > > > > > > > > [name;tld];;
> > > > > > > > > > http://cbbrowne.com/info/slony.html
> > > > > > > > > > "Feel free to  contact me (flames about my english  and
> > > > > > > > > the useless of
> > > > > > > > > > this driver will be redirected to /dev/null, oh no, it's
> > > > > > > > > full...)"
> > > > > > > > > > -- Michael Beck, describing the PC-speaker sound device
> > > > > > > > > >
> > > > > > > > > > ---------------------------(end of
> > > > > > > > > broadcast)---------------------------
> > > > > > > > > > TIP 2: Don't 'kill -9' the postmaster
> > > > > > > > > >
> > > > > > > > >
> > > > > > > > >
> > > > > > > > > ---------------------------(end of
> > > > > > > > > broadcast)---------------------------
> > > > > > > > > TIP 5: don't forget to increase your free space map
> > > > > > > > > settings
> > > > > > > > >
> > > > > > > >
> > > > > > > >
> > > > > > >
> > > > > >
> > > > >
> > > >
> > >
> >
>

In response to

Responses

pgsql-general by date

Next:From: Marc EvansDate: 2006-12-26 15:30:31
Subject: Re: Clustering & Load Balancing & Replication
Previous:From: Devrim GUNDUZDate: 2006-12-26 15:17:30
Subject: Re: Clustering & Load Balancing & Replication

Privacy Policy | About PostgreSQL
Copyright © 1996-2014 The PostgreSQL Global Development Group