Re: redo per second (size)on exadata ?
Date: Tue, 2 Sep 2014 23:13:25 -0500
Message-ID: <CAKb+SBVH-ip-vw1tAeGS2-fkqJgY4dMPjjpLKB72ipO5orhT_g_at_mail.gmail.com>
You're correct. It's 80MB. I have to watch better on the automatic speller in the Android phone
On 3 Sep 2014 00:15, "Seth Miller" <sethmiller.sm_at_gmail.com> wrote:
> Amihay,
>
> Just to clarify and as Riyaj has already mentioned, you are talking about
> "megabytes per second" and not "megabits per second", correct? There is a
> very big difference (a factor of 8).
>
> If you are talking about megabytes, it would be written as 80MB, not 80Mb.
>
> Seth Miller
>
>
>
>
> On Tue, Sep 2, 2014 at 3:47 PM, amihay gonen <agonenil_at_gmail.com> wrote:
>
>> Hi David,
>> thanks for your answer and for other answers and suggestions . it is
>> really help and this what make this forum an excellent forum.
>>
>> I'll try to provide more input. Yes , we are talking on 80Mb per second
>> of redo ,actually I've able to squeeze even then 100mb of redo per second
>> per node from our exadata (quarter machine ).
>>
>> The reason for it is that we are currently checking the limits of our
>> product (http://www.axxana.com/ , sorry for the PR :-) ). Our
>> product integrate into exadata (or any another oracle ) and provide
>> no-data loss protection .
>>
>> my question that was raised before is what will the redo per second
>> generating of high-end oracle (OLTP) setup , or in other words what should
>> be scale the our product (software+hardware) should support.
>>
>> Thanks,
>> Amihay
>>
>> p.s. If anyone want more input on our solution , feel free to send me
>> message in private
>>
>>
>>
>>
>>
>> On Tue, Sep 2, 2014 at 8:50 PM, David Fitzjarrell <
>> dmarc-noreply_at_freelists.org> wrote:
>>
>>> If she really means 80 mb per second on the redo generation that would
>>> generated a large number of log switches; presuming you are using the
>>> recommended redo log size of 4gb we're talking one log switch every 50
>>> seconds. You're not restricted to 4 gb logs so if this is anticipated to
>>> be the 'normal' traffic on your OLTP system you might want to consider
>>> even larger logs. My guess on this is the 80 mb per second isn't the redo
>>> rate, it's the transaction rate and that could mean less actual redo volume
>>> generated. You really need to clarify what this 80 mb per second rate
>>> really means.
>>>
>>> David Fitzjarrell
>>>
>>> Principal author, "Oracle Exadata Survival Guide"
>>>
>>>
>>> On Thursday, August 28, 2014 10:06 AM, "Powell, Mark" <
>>> mark.powell2_at_hp.com> wrote:
>>>
>>>
>>> I imagine that the redo rate varies greatly between systems. How you
>>> looked to see what your past rate or redo generation happens to be? You
>>> ought to be able to use that as a comparison number or baseline. My main
>>> system appears to only generate between 30G – 40G of redo per day across
>>> both nodes. I have several other systems that generate far less.
>>>
>>>
>>> *From:* oracle-l-bounce_at_freelists.org [mailto:
>>> oracle-l-bounce_at_freelists.org] *On Behalf Of *amihay gonen
>>> *Sent:* Thursday, August 28, 2014 6:55 AM
>>> *To:* ORACLE-L
>>> *Subject:* redo per second (size)on exadata ?
>>>
>>> Hi all,
>>> I've been asked by our VP to give estimation what is consider heavy
>>> system OLTP in term of redo per bytes rate.
>>>
>>> She told to to test our exadata machine with load of 80Mb per second
>>> per Node , and I've told her that I think it is too much .
>>>
>>> if OLTP system with generate 80Mb* (2 nodes) per second that it means
>>> 576G per hour .
>>>
>>>
>>> I wonder if anyone work with such systems , what is the typical redo
>>> rate ?
>>>
>>>
>>> thanks
>>> amihay
>>>
>>>
>>>
>>>
>>>
>>>
>>
>
-- http://www.freelists.org/webpage/oracle-lReceived on Wed Sep 03 2014 - 06:13:25 CEST