Migration from EVA to DMX-3 Storage

here the case is bit difficult as you have mutipple vendor storages (HP to EMC)...

The best solutions will be any of host base replication softwares ...

I have experienced on IBM softek software , which i feel very gud one & we have finished 30+ production nodes ..

Gudluck

Prasanth.

You can use EMC Open Replicator for this purpose.Open Replicator migrates data between Symmetrix DMX arrays and third-party storage arrays
within the SAN infrastructure without interfering with host applications and ongoing business operations.

For MSCS environment You can check document at: http://powerlink.emc.com/km/live1/en_US/Offering_Technical/Technical_Documentation/300-006-127.pdf?mtcs=ZXZlbnRUeXBlPUttQ2xpY2tTZWFyY2hSZXN1bHRzRXZlbnQsZG9jdW1lbnRJZD0wOTAxNDA2NjgwNDJlYzA5LGRhdGFTb3VyY2U9RENUTV9lbl9VU18w

You can also check EMC Solution Enabler Symmetrix OPEN Replicator CLI guide at :http://powerlink.emc.com/km/live1/en_US/Offering_Technical/Technical_Documentation/300-002-182_a05_elccnt_0.pdf?mtcs=ZXZlbnRUeXBlPUttQ2xpY2tTZWFyY2hSZXN1bHRzRXZlbnQsZG9jdW1lbnRJZD0wOTAxNDA2NjgwMzEyZDdlLGRhdGFTb3VyY2U9RENUTV9lbl9VU18w

Amita

Please be advised that, if you use OpenReplicator from third party array to EMC Symmetrix/V-Max, you will have to perform a full copy and the application has to be shut down during the copy. So if you decide to copy 10 TB, your downtime will be substantial. IMHO, the FAs doing the data pull should not be used for regular host access.

I disagree with Uwe's comments. The Hot Pull function of Open Replicator can be used for this migration while Host remain active. The only outage would be for the cut-over to the DMX when migration has fully sync'ed. You should possibly use the throttle function to prevent impact to host during migration.

It might be worth expoloring PowerPath Migration Enabler, host copy technology. The data transfer can be done live as well as the switch over. The only reboot required is when installing PP/ME. I checked the documentation and it seems that there are a few EVA models supported.

https://powerlink.emc.com/nsepn/webapps/btg548664833igtcuup4826/km/live1/en_US/Offering_Technical/Technical_Documentation/300-008-343.pdf?mtcs=ZXZlbnRUeXBlPUttQ2xpY2tDb250ZW50RXZlbnQsZG9jdW1lbnRJZD0wOTAxNDA2NjgwNGM3ZTJiLG5hdmVOb2RlPVNvZndhcmVEb3dubG9hZHMtMw__

Good luck!

-Pete

Booyah,

Regarding your comment about the hot pull option of Replicator, I have copied the relevant lines out of the OpenReplicator manual for you.

◆ On pull operations, the remote devices should not be updated by array hosts for the duration of the copy process.

....

◆ Data corruption to devices may be possible during a copy operation if another
host on the SAN has write access to the remote device. If this is a concern, EMC
recommends that the remote device be unmounted or marked as Not Ready to
any other hosts on the SAN to guarantee that the device cannot change while
copying is in process.

...

For online (-hot) copying to the DMX, the control devices may be Read/Write
enabled. The remote devices should not be receiving any updates from their local host.

With the copy operation complete, you can (if necessary) restart the remote
application on the remote host. However, any changes to remote data at this point are
not migrated to the DMX array unless you perform another full Open Replicator pull
operation.

UWE,

These bullets are correct that you point out, but as I understand the product, when it references the remote device (in this case), it is referring to the target device which would be the DMX where the customer would be migrating to. Of course you would not want any host at the remote array to try to access the new remote devices until you have completed the migration and cut the servers over to that new array. Keep in mind we are referring to a data migration. So the bullets you are pointing out are merely stating not to give access to the new remote devices (to any host) until you have completely migrated your data and cut your servers over to the new array. For a Hot Pull operation, your production host continues to use the local EVA devices while the migration takes place and the DMX pulls the data from the EVA to the DMX. From the host's point of view, it will be accessing the EVA devices while the migration takes place.

The challenge with PPME is that for clusters EMC recommends only leaving one cluster node online during the migration. This leaves the cluster at risk during the migration and if running active/active clusters (or clusters with multiple active nodes and one passive node) there may not be enough resources to run all application instances on one node for the duration of the migration.

petehodges wrote: It might be worth expoloring PowerPath Migration Enabler, host copy technology. The data transfer can be done live as well as the switch over. The only reboot required is when installing PP/ME. I checked the documentation and it seems that there are a few EVA models supported.

UWE,

These bullets are correct that you point out.

After reviewing the product manual, I recognize that I have confused some of OR's terms with SRDF terms. Let me rephrase my solution statements.

When migrating from a third party array to a DMX, it can be done online while the data migration takes place. There would be a host down time to cut-over (channel connect) to the new DMX. This cut-over would actually be prior to the migration rather than after as I incorrectly stated before.

Proper terminology = New DMX devices are the control devices. Old storage devices are the remote devices.

The OR session would be started on the new DMX and this DMX would service host I/O while pulling data from the old storage array.

In Pull mode, the Symmetrix volume can again be in a live state during the copy process, which makes migrating from other storage platforms very fast and efficient. The local hosts and applications can begin to access the data as soon as the session begins, even before all of the data has copied over.

Wht about Open Migrator..?

needs less distruption & require downtime during the product installtion & cut overing....

Hi ,

We are using OM3.11 and PPME for one of the Data Migration Project.

We are migrating from HP EVA's 5000/8000 and Calrrions to DMX /V-Max .

Servers include 2003, 2008 ,Window CORE, Hyper-V ,MSCS Clusters as well.

-Both OM and PPME are hostbased applications which require host cpu cycles as we all know.

-Coming to downtime

- OM requires 2 downtimes

(one for installation,attaching drivers and kickoff,

second one for Cutover(to move to Targets)

-PPME just needs a single downtime during initial installation of Powerpath stuff

FYI: Please check OM3.12 documentation as i heard that the way it works with MSCS clusters have changed (i think we need not shut down the passive node (please check as i have not checked yet)

Powerlink documentation should be good enough for the migration,however please feel free to contact me for any specific scenarios you want to discuss .

Regards

Pawan

malisetty_pawan@emc.com

Message was edited by: Pawan_M