Thanks guys for your replies. I really appreaciate your help and advices. I will look at the mentioned products in your answers and possible solutions for me.

Hello Ernes,

reg. your answer....."a) Yes. Using MirrorView", what transport protocol is used between sites?

Thanks,

Hejbi

Thank you all of guys, I created 3 potential scenarios, what do you think, which is the best one? (My answers to Rakesh - RPO, RTO from hours to weeks; Applications - all kinds of application (db servers, app servers, file servers,....); Some data is critical, some less critical; no ISL connection, only 1x director clusters in each DC; OS - Linux, Win, Unix; used VMware ESXi, x86 physical servers + AIX). I appreciate all comments. Thanks. Hejbi

Your life will be a lot simpler with RecoverPoint. You can use AppSync with RecoverPoint to create application consistent PITs (Point In Time) in RP.

a) Yes. Using MirrorView

b) Depends on the latency. 2 SANs will most likely be your answer

c) To fully answer this question you would need to look at the latency between the sites and the amount of writes on the source. The answer is it depends again. Have I seen it before? Yes.

d) Local snaps? Yes. You can use RM or AppSync. AppSync is the better choice as long as the app is supported by AppSync

You can try MirrorView: http://www.emc.com/collateral/hardware/white-papers/h2417-mirrorview-know-cx-series-flare-wp-ldv.pdf. The distance is 500km so maybe only MirrorView/A can work in your environment.

Hi Hejbi,

You will have to look on multiple questions :

What is RPO and RTO ?

what is budgeting ?

Which Applications are running ?

How much data is critical ?

What about IOPS ?

What is ROI target ?

Which Switch and I/O modules (on storage) you will use ?

Will there be any ISL connections, if yes how many ?

What OS and RAID types ?

Both DCs will be Active/Passive or Active/Active ? etc.

Answers to your questions :

a) You can use MirrorView/A. It works with FC & iSCSI, no additional hardware is needed. But it site 2 will be standby you need to promote a MirrorView  and redirect users to it. The other option is vRPA (Virtual RecoverPoint Appliance), its s/w solution, needs standard ESX hardware.

b) It will depend on RPO and RTO. Both ways are okay, again you want it to work as Active/Standby or Active/Active.

c) Yes, it supports.

d) Yes, VNX supports consistent snapshot with Replication manager. You don't need other tool. RM itself contacts to Applications (like MS Exchange and SQL) before a snapshot for consistency.

Thnx

Rakesh

b) if you are going to use array based replication technologies, do not extend fabrics between two location, especially at this great distance. Keep fabrics local to physical location and use technologies such as Cisco IVR/Transit VSAN to establish connectivity without merging two fabrics.

All the answers here are on point. I wanted to ask if the arrays are going to be hosting virtual machines, and if so, will all the required "protected" systems be virtual (with VMware, assumedly)?

If so, would you be planning to use VMware SRM?

Best,

Steve

Ports should be FC & iSCSI. Along with them you can use DWDM and FCIP. So all 4 are supported, depending your environment and distance with need.

Looking at your first post, if I have to choose, I will go for 2nd option. It's more reliable and faster than the 3rd one and cost effective than the 1st one.

Thanx

Rakesh