Go Premium for a chance to win a PS4. Enter to Win

x
  • Status: Solved
  • Priority: Medium
  • Security: Public
  • Views: 2614
  • Last Modified:

EMC Recover Point or MirrorView/A

Environment will be Blade Servers running VMWare w/ CLARiiON CX-3 20. Cookie cut at DR site. Goal is to sync the LUNs on the two arrays between sites (will be over a WAN link so synchronous is out) and use VMWare Site Recovery Manager to automate a recovery in the event of a failure.
I understand Recover Point is probably the ideal solution but for cost justification and feasibility, would MirrorView/A be a sound technology for what I want to accomplish? I am comfortable with a 15 minute update interval. Also will MirrorView/A be solid with Exchange databases for consistency?
0
ChopperCentury
Asked:
ChopperCentury
  • 6
  • 4
1 Solution
 
Duncan MeyersCommented:
MirrorView/A and RecoverPoint/SE will both provide replication - however there are some big differences. MirrorView/A can replicate across iSCSI and but cannot provide application consistency for Exchange - a MirrorView/A replicated copy will be crash-consistent at best. You can take a clone of a replicated copy at the DR site and do some fancy scripting to mount the clone on a host and run ESEUTIL against the Exchange databases - but at the moment, there's no automated process to do that. EMC's Replication Manager can provide that level of automation, but not with MirrorView/A yet. Another option is to use Replication Manager to automate the process of taking clones/snaps and then replicate them using SANCopy as an altenative to RecoverPoint and MirrorView/A. The other thing to note is that the 15 minute replication gap means a minimum of a *30 minute* RPO gap.

RecoverPoint has some BIG advantages - it can get about 6:1 compression on WAN traffic, so you don't need as big a link as you would with MirrorView/A or, alternatively, you can get much better RPOs or protect more servers with the bandwidth you've got. You can also take 'bookmarks' with RecoverPoint. YOu'd typically use Replication Manager handle the work of queiescing a server (Exchange, SQL, Oracle etc), flush buffers, then insert a bookmark into the RecoverPoint journal - and taht bookmark represents a gold copy (100% consistent and recoverable) of the database. RecoverPoint also gives you the ability to roll back one write operation at a time so you could easily recover from, say, a zero-day malware outbreak quickly and easily. RecoverPoint intergrates easily with CLARiiONs as teh CX3- and CX4- storage systems have teh RecoverPOint write splitter built into the CLARiiON and it just has to be switched on. The downside of RecoverPoint/SE is that it is licensed by protected storage (4TB, 8TB, 16TB IIRC), and you need additional fast disk for the RecoverPoint journals - and the amount of journal disk you have determines how far you can roll back. It is also more expensive than MirrorView - but you save so much on WAN bandwidth costs over other replication mechanisms that it typically pays for itself in 12 to 18 months.

I like RecoverPoint. It rocks!
0
 
Duncan MeyersCommented:
Hi ChopperCentury,

Did you have any other questions?
0
 
ChopperCenturyAuthor Commented:
I realize Recover Point is probably the best but cost prohibitive.
By using MirrorView /A and SRM from VMWare, I have read that I can recover Exchange, SQL, etc... Only that the data will be from the last succesful snapshot MirrorView passed across.
All I want is to be able to recover application data in a consistent state; will probably have a 10 to 15 minute update interval. WAN link is about 20Mb. Would this work or is Recover Point required and Replication Manager?
thanks
0
Technology Partners: We Want Your Opinion!

We value your feedback.

Take our survey and automatically be enter to win anyone of the following:
Yeti Cooler, Amazon eGift Card, and Movie eGift Card!

 
Duncan MeyersCommented:
How much data do you want to protect with the replication solution and what is the rate of change?
0
 
ChopperCenturyAuthor Commented:
Will be replicating all VM's. All existing WinTel servers will be migrated to VM's in a blade environment. Will replicate those LUNs. Servers will be domain controller, email, database, file, print, web, file servers, etc... roughly 30 to 40 servers. These are not very high traffic systems though, the rate of change will not be too heavy.
0
 
Duncan MeyersCommented:
OK - how many GB do you back up in a full weekly backup, and how much data do you back up in a daily incremental backup?
0
 
ChopperCenturyAuthor Commented:
Roughly 2 to 3GB daily and around 1TB for full
0
 
Duncan MeyersCommented:
Are you sure? That's a rate of change of about 0.2%. Typical rates of change are around 3% at file level and 1.5% at block level, so I'd expect for 1TB of full backups, 30GB of daily incrementals or thereabouts - more if you are doing agent-based backups of SQL Server or Exchange.
0
 
ChopperCenturyAuthor Commented:
Decided to hold of on making a firm decision until the VMs are in place and can see exactly how much data will need to be replicated. Thanks for the detailed information.
0
 
Duncan MeyersCommented:
No worries. Thanks for the points!

Bear in mind that you'll be able to replicate a heap more across your existing link using RecoverPoint than MirrorView, so factor in cost of link upgrades when you make your decision. As a starting point:
20Mb/sec = 2MB/sec
2MB/sec = 7.2GB/hr

Assuming 30GB per day rate of change, and that rate of change will happen in production hours (assume 8 hour day)
30GB/8 hrs = 3.75GB/hr

You can replicate 7.2GB/hr if the link is dedicated to replication traffic, so you could have an asynchronous replication period of about 30 minutes, or RPO of 1 hour using MirrorView/A. For RecoverPoint, assume a 6:1 compression ratio on on the WAN link:
2MB/sec * 6 = 12MB/sec
12MB/sec = 43GB/hr

Assuming 30GB per day rate of change, and that rate of change will happen in production hours (assume 8 hour day)
30GB/8 hrs = 3.75GB/hr

Or an RPO of about 5 minutes. Or you can reduce the link bandwidth or you won't need to upgrade the link if you are putting a mix of replication and network traffic across the same link. RecoverPoint turns you 20Mb/sec link into an effective 120Mb/sec link - so you can see that RecoverPoint can make good economic sense as well as technology sense.

Note that RecoverPoint can achieve over 10:1 WAN compression, but the appliances are getting bogged down at that sort of compression ratio. 6:1 is a good place - it gives you excellent link utilisation while leaving plenty of processor headroom on the RecoverPoint appliances.
0

Featured Post

What is SQL Server and how does it work?

The purpose of this paper is to provide you background on SQL Server. It’s your self-study guide for learning fundamentals. It includes both the history of SQL and its technical basics. Concepts and definitions will form the solid foundation of your future DBA expertise.

  • 6
  • 4
Tackle projects and never again get stuck behind a technical roadblock.
Join Now