In the familiar SnapView Snapshot environment, the COFW process reads the original 64 KB data chunk from the source LUN, writes that chunk to the Reserved LUN, and updates the pointers in the Reserved LUN map area. Once these steps complete, the host write to the Source LUN is allowed to proceed, and the host will receive an acknowledgement that the write is complete. If a SnapView Snapshot is deleted, data in the RLP is simply removed, and no processing takes place on the Source LUN.
In the case of a VNX Snapshot, a new host write is simply written to a new location (redirected) inside the Pool. The original data remains where it is, and is untouched by the ROW process. The granularity of Thin LUNs is 8 KB, and this is the granularity used for VNX Snapshots. New data written to a Thick LUN with a VNX Snapshot is mapped in the same way as the data for a Thin LUN, and it is expected that this will reduce Thick LUN performance. If a VNX Snapshot is removed from a Thin LUN, data will be consolidated into 256 MB slices to minimize wasted space. If the last VNX Snapshot is removed from a Thick LUN, the defragmentation process moves the new data to the original locations on disk, and freed space is returned to the Pool.
In this slide, the processes involved in a secondary host read of a Snapshot are compared. In the familiar SnapView Snapshot environment, data which has not yet been modified is read from the source LUN, while data that has been modified since the start of the SnapView Session is read from the Reserved LUN. SnapView always needs to perform a lookup to determine whether data is on the Source LUN or Reserved LUN, which causes Snapshot reads to be slower than Source LUN reads.
In the case of a VNX Snapshot, the original data remains where it is, and is therefore read from the original location on the Primary LUN. That location will be discovered by a lookup which is no different to that performed on a Thin LUN which does not have a VNX Snapshot, so the performance is largely unchanged.
SnapView snapshots are a view of the data and not the actual data itself, it is an instantaneous frozen virtual copy of a LUN on a storage system. A SnapView snapshot is a composite of the unchanged data chunks on the source LUN and data chunks on a LUN called “Reserved LUN”. Before chunks of data are written to the source LUN, they are copied to a reserved area in private space, and the memory map is updated with the new location of these chunks. This is referred to as COFW. It uses pointers to track whether data on the source LUN, or in the Reserved LUN Pool.
Thanks for the details. If we have Pool on VNX, and we would like to take snapshots - Local Data protection, should we get Snapshot license or SnapView license? This is what I am trying to understand to use these products at the right place. I understand, if we want to clone LUN, we definitely need to get SnapView License.
You get all local snapshot capabilities with the Local Protection Suite so you will get both VNX Snapshots, SnapView, as well as RecoverPoint/SE CDP and SnapSure (for file if the VNX is Unified).
For snapshots, it really depends. Most new installations these days will probably use VNX snapshots as there are advantages to using them (more snaps per LUN supported, no copy on first write, no separate LUN pools, snaps of snaps, etc.). They do require pools and do not work with classic LUNs. If you are not using pools, then you will need to use SnapView because that is what is supported.
With respect to cloning, it depends on what you need. A writable snap could serve the purpose of a clone.
There are very good white papers and manuals out there that can show syntax, etc.
Zhang_Jiawen
2 Intern
•
1161 Posts
2161
0
Posted May 28th, 2014 20:00
In the familiar SnapView Snapshot environment, the COFW process reads the original 64 KB data chunk from the source LUN, writes that chunk to the Reserved LUN, and updates the pointers in the Reserved LUN map area. Once these steps complete, the host write to the Source LUN is allowed to proceed, and the host will receive an acknowledgement that the write is complete. If a SnapView Snapshot is deleted, data in the RLP is simply removed, and no processing takes place on the Source LUN.
In the case of a VNX Snapshot, a new host write is simply written to a new location (redirected) inside the Pool. The original data remains where it is, and is untouched by the ROW process. The granularity of Thin LUNs is 8 KB, and this is the granularity used for VNX Snapshots. New data written to a Thick LUN with a VNX Snapshot is mapped in the same way as the data for a Thin LUN, and it is expected that this will reduce Thick LUN performance. If a VNX Snapshot is removed from a Thin LUN, data will be consolidated into 256 MB slices to minimize wasted space. If the last VNX Snapshot is removed from a Thick LUN, the defragmentation process moves the new data to the original locations on disk, and freed space is returned to the Pool.
In this slide, the processes involved in a secondary host read of a Snapshot are compared. In the familiar SnapView Snapshot environment, data which has not yet been modified is read from the source LUN, while data that has been modified since the start of the SnapView Session is read from the Reserved LUN. SnapView always needs to perform a lookup to determine whether data is on the Source LUN or Reserved LUN, which causes Snapshot reads to be slower than Source LUN reads.
In the case of a VNX Snapshot, the original data remains where it is, and is therefore read from the original location on the Primary LUN. That location will be discovered by a lookup which is no different to that performed on a Thin LUN which does not have a VNX Snapshot, so the performance is largely unchanged.