This post is more than 5 years old
2.2K Posts
0
2704
Mirror Positions available on R5 STD?
I am trying to figure out what my options are for associating additional BCVs for some STD R5 devices on my DMX-4 running 5773. I currently use Replication Manager to sync a BCV nightly to the STD production devices and then create TF/Snaps of the BCV. I now have a requirement for some additional TF/Snaps that will be on a different refresh schedule and so need to add another BCV to accommodate that new schedule. As I understand it there are 4 mirror positions available on a STD device and that R5 uses 2 of those positions (by reading the output below). It is also my understanding that a BCV will occupy a third mirror position, but only while established and that the mirror position is released once split. I also have Optimizer that is scheduled to perform swaps on a different schedule than the BCV synchronization to avoid conflicts. And Optimzer will occupy a mirror position as well during its moves but releases that mirror position once the move is complete.
So if all my assumptions are correct I can add another BCV as long as my schedules are such that the BCVs and Optimizer are not running at the same time and therefore exceeding the 4 mirror positions?
Below is the relevant output from one of the devices that need the additional BCV.
# symdev show 2e4
Device Configuration : RAID-5 (Meta Head,
Non-Exclusive Access)
Mirror Set Type : [RAID-5,RAID-5,N/A,N/A]
Mirror Set DA Status : [RW,RW,N/A,N/A]
Mirror Set Inv. Tracks : [0,0,0,0]
BCV Pair Information
{
Standard (STD) Device Symmetrix Name : 02E4
Standard (STD) Device Serial ID : Not Visible
Standard (STD) Device Group Name : mcprod
Standard (STD) Composite Group Name : Not/Grouped
BCV Device Symmetrix Name : 00EF
BCV Device Serial ID : Not Visible
BCV Device Associated Group Name : MCProd_BCV1
BCV Device Associated CG Name : Not/Associated
BCV Device Status : Not Ready (NR)
State of Pair ( STD <=\=> BCV ) : Split
Time of Last BCV Action : Thu Aug 12 00:13:26 2010
State of BCV Mirrors : Synchronized
BCV State Flags : (AllReady)(Emulation)
Percent Split : 100%
Number of Inv. Tracks for STD Device : 0
Number of Inv. Tracks for BCV Device : 0
}
Quincy561
1.3K Posts
0
August 13th, 2010 08:00
True BCVs do use one or two mirror postions when established, but clones do not. RAID5 BCVs are always really clones which do not use mirror positions.
Booyah2
184 Posts
0
August 12th, 2010 13:00
You should look into using TimeFinder / Clone copies for the BCV config. The Clone does not use a mirror position.
In fact, you can have up to 16 clones against one source, with automatic background copies to 8 targets simultaneously.
Hope this helps...
If this is helpful or answers your question, please mark reply appropriately, this helps others find relevant answers to thier queries.
Anonymous
5 Practitioner
5 Practitioner
•
274.2K Posts
1
August 12th, 2010 13:00
You can do what you described. If talking to TimeFinder/Snap, clone operations are not really a concern with mirror position. Another way, and maybe a better way, to be handle this would be to open a chat with the software group, specifying whether you are using Windows or UNIX.
Booyah2
184 Posts
0
August 12th, 2010 13:00
If you have a BCV occupying a mirror position, then you split that BCV and then you allow a second BCV to use that same mirror position, then a full establish is required for the first BCV if you were planning to re-establish it.
The problem is that the array has to maintain it's track differential tables for the first BCV, if you allow a DRV or a Hot Spare, etc... to get possession of that BCV mirror slot while the BCV is split, then you loose all your differential tables and a full establish is required for the first BCV.
This is another reason that makes TineFinder Clones so appealing.
AranH1
2.2K Posts
0
August 12th, 2010 16:00
The reason that I did not look into using TF/Clone as the source for the snap is that I use RM to manage the jobs and RM would not allow you to take a snap of a TF/Clone, it had to be a BCV created with TF/Mirror. I will have to find out if that limitation still exists.
AranH1
2.2K Posts
0
August 12th, 2010 16:00
Wow, I did not realize this. So having concurrent BCVs means a full restablish each time? That would not work with us as the devices in question are for a SQL database that is 6TB of storage. It takes too long to make a full establish.
JoeO2
34 Posts
0
August 12th, 2010 18:00
Multi-BCV sync\split only requires an initial full sync\split to pair the stds with the bcvs then incremental sync\split and not a full sync\split thereafter.
AranH1
2.2K Posts
0
August 13th, 2010 07:00
Joe,
What about what Booyah wrote about mirror positions and the result of a mirror position being used by another BCV?
Booyah2
184 Posts
0
August 13th, 2010 08:00
RAID5 in a TimeFinder/Mirror environment still uses a mirror position, but true that it is using the Clone emulation under the covers.
JoeO2
34 Posts
1
August 13th, 2010 08:00
Aran,
Do not think of it as mirror positions (mirrors are only for mirrored devices) but copy sessions.In a RAID5 environment TF clone is limited to 16 copy sessions while TF mirror (which is actually running in clone emulation mode) is limited to 16 - (2 per BCV) copy sessions.Here is an excerpt from TF CLI 7.1 product guide.
The Symmetrix array is currently limited to 16 sessions per source device, which can
be used for TimeFinder/Clone, TimeFinder/Snap, SRDF/Star, Solutions Enabler
Open Replicator (ORS), or Symmetrix Differential Data Facility (SDDF) operations.
This limits the number of available clone copies that can be created. For example:
◆ When a standard device has two BCV device pairs, you can only create up to 14
copy sessions for that standard device.
◆ When TimeFinder/Clone is running in emulation mode, this number is
decremented an additional two sessions per single BCV relationship.
◆ Copy sessions created using the -copy option to create full data copies are limited
to eight concurrent sessions, with up to an additional eight CopyOnAccess
sessions, for a total of 16 sessions.
◆ Copy sessions created with the -differential option uses two session slots.
I hope this clarifies the answers you've received?
Quincy561
1.3K Posts
0
August 13th, 2010 08:00
Just to clarify,a RAID5 BCVs will not take a mirror position on the source device like a true mirror BCV will when established because it is not a true BCV but is a clone. A clone when established will not take a mirror position from the source device.
Quincy561
1.3K Posts
0
August 13th, 2010 08:00
RAID5 BCV targets are always clones under the covers and will never use mirror positions, even if you use the emulation commands (symmir) to establish them. It might make sense to switch to native clone commands to avoid confusion.
AranH1
2.2K Posts
0
August 13th, 2010 08:00
Alright so I think all three of you are in agreement now?
RAID5 BCVs will NOT occupy a mirror position and I do NOT have to be concerned about needing to perform a full establish due to overcommitment of mirror positions when establishing multiple RAID5 BCVs to the source devices?
Booyah2
184 Posts
0
August 13th, 2010 08:00
I agree
And if you just want to get a better comfort level, just do a symdev show on your device you listed in the origianl post while established and check the "mirror set type"
The N/A,NA, below are the other two available mirror positions for device 2e4.
# symdev show 2e4
Device Configuration : RAID-5 (Meta Head,
Non-Exclusive Access)
Mirror Set Type : [RAID-5,RAID-5,N/A,N/A]
AranH1
2.2K Posts
0
August 13th, 2010 08:00
So even though my RM Job is configured to use TimeFinder/Mirror and the RAID5 devices that are used in the job show as BCV+R5 when I perform a 'symdev show' on them, the symm is actually treating them like a clone and those devices will not occupy a mirror position?