Hi Philippe,
To put your numbers in perspective, the base config with 1 DB - 1
controler - SingleDB - NO recoverylog achieves 100 inserts in 813ms,
which is 0.8ms per insert.
When you put 2 DB - 2 controlers - RAIDb-1 - 2 RecoveryLog -
Distribution, the average request execution time becomes 2746ms.
This basically mean that Sequoia is roughly adding a 2ms latency to your
requests when they are executed by all nodes in your cluster (including
network round-trips). As your benchmark is purely sequential you pay
that latency on every single request and this is why you see the time
increasing.
Sequoia does not make any single request faster, which means that your
database is not going to execute requests faster because Sequoia is
there. Actually Sequoia is going to add some overhead to every single
request.
However, Sequoia can provide scalability, which means that when your
volume of requests increases (more requests in parallel), it will be
able to distribute the load evenly on your backends. Sequoia does not
implement a parallel database where you have intra-request parallelism
(we don't split a request in small chunks and distribute it), but it
provides inter-request parallelism (different requests are routed to
different backends). Therefore you can't see the benefits of Sequoia
with a single threaded benchmark but you can measure the latency that
will be added to individual requests.
Regarding the network traffic, there are multiple Appia configurations
that are possible. You can use TCP or UDP, and there are several
algorithms on how the total order token can be managed. Some
configurations where the token is greedily passed between the members of
the group will generate a heavy network traffic. You can experience with
other configuration and also try JGroups.
Thanks for your interest in Sequoia,
Emmanuel
Versions: sequoia 2.10.10
Environment: 2 nodes - Linux
2 controlers Sequoia 2.10.10
2 BDD mysql on the same node that Sequoia
2 ReceveryLog mysql on the same node that Sequoia
Config RAIDb-1
Appia default config Network with Go full duplex router between controlers
Hello
With this config, the insert performance are very bad. On the whole an
insert take 3 times that an insert directly in base.
Table in BDD: create table benchSequoia01 (col1 int, col2 int, col3
varchar(16))
Requet insert: insert into benchSequoia01 values (val1,val2,'AAA')
1000 insert give this time on different configuration:
1 BDD - 1 controler - SingleDB - NO recoverylog
Boucle d'insert
(1000) [813 ms]
1 BDD - 1 controler - SingleDB - RecoveryLog
Boucle d'insert
(1000) [800 ms]
1 BDD - 1 controler - RAIDb-1 - RecoveryLog - Distribution no
enable in xml file
Boucle d'insert
(1000) [1122 ms]
1 BDD - 1 controler - RAIDb-1 - Recovery - Distribution enable
in xml file
Boucle d'insert
(1000) [1644 ms]
2 BDD - 2 controlers - RAIDb-1 - 2 RecoveryLog - Distribution
enable in xml file
Boucle d'insert
(1000) [2746ms]
The network traffic is very important just for a small insert. 4Mo in
point between les 2 controllers.
Please excuse me for my very bad Inglish.
Thanks for your help
*Philippe PAVY*
Ingénieur Base de données - Atos Worldline
/ Z.I. Rue de la Pointe/
/ 59113 Seclin/
[email protected] <mailto:[email protected]>_
Tel: 33+ (0)3 20 60 *7763* / Fax: 33+ (0)3 20 60 7649
Atos Worldline is an Atos Origin company :
_<http://www.atosorigin.com/>_
*---------------------------------------------------*
------------------------------------------------------------------------
Ce message et les pièces jointes sont confidentiels et réservés à
l'usage exclusif de ses destinataires. Il peut également être protégé
par le secret professionnel. Si vous recevez ce message par erreur,
merci d'en avertir immédiatement l'expéditeur et de le détruire.
L'intégrité du message ne pouvant être assurée sur Internet, la
responsabilité du groupe Atos Origin ne pourra être recherchée quant
au contenu de ce message. Bien que les meilleurs efforts soient faits
pour maintenir cette transmission exempte de tout virus, l'expéditeur
ne donne aucune garantie à cet égard et sa responsabilité ne saurait
être recherchée pour tout dommage résultant d'un virus transmis.
This e-mail and the documents attached are confidential and intended
solely for the addressee; it may also be privileged. If you receive
this e-mail in error, please notify the sender immediately and destroy
it. As its integrity cannot be secured on the Internet, the Atos
Origin group liability cannot be triggered for the message content.
Although the sender endeavours to maintain a computer virus-free
network, the sender does not warrant that this transmission is
virus-free and will not be liable for any damages resulting from any
virus transmitted.
------------------------------------------------------------------------
_______________________________________________
Sequoia mailing list
[email protected]
https://forge.continuent.org/mailman/listinfo/sequoia
--
Emmanuel Cecchet
FTO @ Frog Thinker
Open Source Development & Consulting
--
Web: http://www.frogthinker.org
email: [email protected]
Skype: emmanuel_cecchet
_______________________________________________
Sequoia mailing list
[email protected]
https://forge.continuent.org/mailman/listinfo/sequoia
