chazzyd |
Storage based. You don’t want to worry about identical servers, support, and all that.
When it comes to disaster recovery, the data is the key consideration.
chazzyd |
Storage based. You don’t want to worry about identical servers, support, and all that.
When it comes to disaster recovery, the data is the key consideration.
epeterson |
Storage based. It moves the replication function to a level that makes what’s going on at the host level independent of the requirement to move date from one place to another. If you ever add another host platform this method will work just as effectively. From a process prespective it solves a lot of problems.
The only issue is that all of the data you want to replicate must me on the storage system. With a host based solution that requirement does not exist.
The only replication method the our storage team supports is host based.
epeterson |
I made a type on my earlier reply. The storage team only supports storage based replication.
odd1atRSA |
Think outside of the boxes…. Think in terms of ease of service delivery in a DR situation. When the IT staff is stressed to the max, by having to recover from a disaster, simplicity is paramount. Could the existing cluster be turned into a campus cluster? Do you own all of the DR planning or just the implementation?
After you define the failure modes that you may need to recover from and their respective recovery scenarios, you should look at the cost of each, remembering requirements are cost drivers and cost is a requirement driver.
Should you take the SC9900 from development where is is probably used for integration testing and pre-production tuning and get them something else? Depends on your production acceptance criteria.
Should you do storage subsystem data replication, host based, or implement an appliance solution? Again it depends on your requirements, project life-cycle costs, and present infrastructure.
Sorry no simple answers, but in general storage administrators like storage based solutions, systems administrators like host based, and network administrators like appliances.
satria2555 |
guys thanks for replying, do you mind adding some more, your role & expertise, your environment, what is your replication based….thanks
DataCenter, Disaster Recovery, IT architecture, EMC, Hitachi Data Systems, Sun Microsystems, Veritas, Backup & recovery, Storage servers
hi, i have 2xSE9900. one for the Production (prod) and one for Development (dev), few clustered DB servers (some inter domain and some normal clustered) which total more than 5. the servers are running different Solaris versions, VXVM versions but all using Sun cluster. for the next phase of implementation the number of DB servers gonna increase. i tot of moving DEV SE9900 to DR site and replace the current DEV data inside the lower end storage box. i have few vendors coming in. appliance-based, host-based and storage-based replication. i dont want appliance-based. so my options is whether host-based or storage-based. host-based needs all the servers to be installed with the same versions. the storage-based requires the DEV SE9900 to be migrated to DR site. from your pro experience, which one is better? easier to implement, manage and operate? the money and bandwidth is not an issue. the data is about 11TB. 
0
