Re: BDR
От | Alvaro Aguayo Garcia-Rada |
---|---|
Тема | Re: BDR |
Дата | |
Msg-id | 1925727693.13007.1465822110294.JavaMail.zimbra@opensysperu.com обсуждение исходный текст |
Ответ на | Re: BDR (Rakesh Kumar <rakeshkumar464a3@gmail.com>) |
Список | pgsql-general |
Block based replication is the replication mechanism postgres incorporates natively. It's, in brief, sending all the file-levelchanges to all the slaves, so the data folder is always the same. It's like having a replicated folder, not includinglogs and some other things. The disadvantage of block level replication, according to BDR, it that, when you have an open transaction, all DML is writtento disk, even when the transaction is not committed. In a large transaction, all the date would be send to the slaves,even when, at the end, the transaction is rolled back. With BDR, the transaction is sent to the other masters onlyonce it's committed. Of course, this can be a problem in both cases, depending on your environment. With block level replication, you can getunnecessary traffic for transactions that would be finally rolled back(in contrast with BDR, which will send the wholetransaction once it's committed); on BDR, you will get traffic peaks for some large transactions, as the whole transactionis sent once it's committed(in contrast to block level replication, which would send changes as they are beingexecuted). The later can also cause some delay, depending on the connection between the servers: Regards, Alvaro Aguayo Jefe de Operaciones Open Comb Systems E.I.R.L. Oficina: (+51-1) 3377813 | RPM: #034252 / (+51) 995540103 | RPC: (+51) 954183248 Website: www.ocs.pe ----- Original Message ----- From: "Rakesh Kumar" <rakeshkumar464a3@gmail.com> Cc: "PostgreSql-general" <pgsql-general@postgresql.org> Sent: Monday, 13 June, 2016 07:13:09 Subject: Re: [GENERAL] BDR http://bdr-project.org/docs/next/logical-vs-physical.html "It (BDR) has significant advantages - and some disadvantages - when compared to PostgreSQL's older physical (block-based) streaming or archive-based replication with warm or hot standby" What exactly is block based? Changes are recorded in the redo log, right? Does that mean that in streaming replication, from redo log the server applies changes at the block level of the data-file. That would also mean that at any time, both primary and standby would be exactly same, block by block. -- Sent via pgsql-general mailing list (pgsql-general@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-general
В списке pgsql-general по дате отправления: