Re: Database migration options

From: Chris Taylor <christopherdtaylor1994_at_gmail.com>
Date: Thu, 22 Mar 2018 11:06:39 -0500
Message-ID: <CAP79kiRMghJbYZqsFdkm0zZF6bX6=tqLtA7waqUpf7dxbaOywQ_at_mail.gmail.com>



Exactly. That gives you some lead time to get everything built and ready for cutover weekend without taking a long down time.

The GGate setup could be simple or complex - all depends on your environment. GGate is not simple so if you're already familiar with using it then it shouldn't be too much of a problem. But if you're new to GGate, then expect a somewhat steep learning curve.

Chris

On Thu, Mar 22, 2018 at 10:08 AM, Chen Zhou <oracle.unknowns_at_gmail.com> wrote:

> Chris,
> I was thinking the same thing, to do that I would set up GoldenGate
> beforehand, instantiate the tables by using Datapump export with select.
> Then import the data into the new database. Start GoldenGate to apply
> changes from that point on.
> Is that what you are suggesting too?
> Thank you,
> Chen
>
> On Thu, Mar 22, 2018 at 6:16 AM, Chris Taylor <
> christopherdtaylor1994_at_gmail.com> wrote:
>
>> Let me explain my point here a bit better. (I don't know the size of
>> your data of course)
>>
>> You can do your datapump using parallel and if you have GoldenGate you
>> use GoldenGate to keep the new data in sync for the tables you've
>> identified.
>> My whole point was about using datapump + Golden Gate. Then when you're
>> ready to cutover to the new database the data will be there and doesn't
>> have to be done on your go-live weekend or whatever. (since you mentioned
>> that Golden Gate might be an option).
>>
>> Chris
>>
>>
>>
>> On Wed, Mar 21, 2018 at 10:47 PM, Chen Zhou <oracle.unknowns_at_gmail.com>
>> wrote:
>>
>>> Chris and Vishnu,
>>> Thank you for your suggestions.
>>> I think Datapump will take at least hours in our case, but it worths a
>>> try to get the timing.
>>> As for partitioning source data to use TTS, I will look into that and
>>> may test it in our lab.
>>> Thank you,
>>> Chen
>>>
>>> On Wed, Mar 21, 2018 at 7:49 PM, Vishnu <vishnukumarmp_at_gmail.com> wrote:
>>>
>>>> Hi Chen, You can try transportable tablespace (TTS) for this. Partition
>>>> your source data and move newer data to new tablespace and transfer it to
>>>> destination. This method would be much faster than datapump export/import
>>>>
>>>> Thanks,
>>>> Vishnu
>>>>
>>>>
>>>> On Wed, Mar 21, 2018 at 10:00 PM, Chris Taylor <
>>>> christopherdtaylor1994_at_gmail.com> wrote:
>>>>
>>>>> Datapump can subset data - you provide a query to DataPump to pull out
>>>>> the data you want to export and import into the new database.
>>>>>
>>>>> You could at least do a POC (Proof of Concept) of datapump to see how
>>>>> what that can get you . You can use parallel on the datapump to speed
>>>>> things up a lot.
>>>>>
>>>>> You could combine Datapump subsetting and GoldenGate I think - subset
>>>>> your data and use Golden Gate to sync up the SCNs (CSNs in GoldenGate) to
>>>>> keep the new data up to date if I'm not terribly mistaken.
>>>>>
>>>>> Chris
>>>>>
>>>>>
>>>>> On Wed, Mar 21, 2018 at 7:50 PM, Chen Zhou <oracle.unknowns_at_gmail.com>
>>>>> wrote:
>>>>>
>>>>>> Hi,
>>>>>> Our company needs to migrate some databases to new servers.
>>>>>> However we don't want to migrate the whole databases, which have
>>>>>> years of old data. Rather, we want to select some newer data from some
>>>>>> tables and only carry these over to the new databases.
>>>>>> Export and import will take too long due to the large sizes of our
>>>>>> databases.
>>>>>> Physical Standby won't work. Is logical standby an option here?
>>>>>> I am also considering using GoldenGate.
>>>>>> Are there other options?
>>>>>> Thank you,
>>>>>> Chen
>>>>>>
>>>>>>
>>>>>
>>>>
>>>
>>
>

--
http://www.freelists.org/webpage/oracle-l
Received on Thu Mar 22 2018 - 17:06:39 CET

Original text of this message