Re: gc buffer busy and rac testing
From: Palooka <nobody_at_nowhere.com>
Date: Thu, 08 Jan 2009 23:22:38 +0000
Message-ID: <3Vv9l.255109$LM6.186623_at_newsfe03.ams2>
Mladen Gogala wrote:
> On Thu, 08 Jan 2009 15:07:19 +0100, helter skelter wrote:
>
>
> Yes, it is. The query is probably a report, accessing rather large amount
> of blocks modified on the other instance. In that case, so called cache
> fusion mechanism will create consistent image of the blocks modified and
> ship them over the cluster interconnect to the node executing the query.
> That takes time. What do you use for the cluster interconnect? I hope it's
> at least 1G ethernet? 10G would be preferred, but at least 1G is
> absolutely a must.
>
>
>
OP said 1G in the first place.
Date: Thu, 08 Jan 2009 23:22:38 +0000
Message-ID: <3Vv9l.255109$LM6.186623_at_newsfe03.ams2>
Mladen Gogala wrote:
> On Thu, 08 Jan 2009 15:07:19 +0100, helter skelter wrote:
>
>> I've done little investigation and I found sql resposible for this >> waits, it's a simple query with join and order, quering two tables. Is >> it normal ?
>
> Yes, it is. The query is probably a report, accessing rather large amount
> of blocks modified on the other instance. In that case, so called cache
> fusion mechanism will create consistent image of the blocks modified and
> ship them over the cluster interconnect to the node executing the query.
> That takes time. What do you use for the cluster interconnect? I hope it's
> at least 1G ethernet? 10G would be preferred, but at least 1G is
> absolutely a must.
>
>
>
OP said 1G in the first place.
Your remarks about RAC being for resilience and availability, rather than performance and scalability, are interesting. No disagreement whatsoever from here, but that isn't how Oracle/Captain Morgan/Sybrand sell it.
Palooka Received on Thu Jan 08 2009 - 17:22:38 CST