Chaps,
I've asked about snapshots before and got some good responses but I think I need to understand the basics a bit more rather than dive straight in... So, hope you dont mind helping me out!
- When using a 3par snapshot, I'm assuming the whole LUN is snapshot to the COPY CPG? As opposed to a VMware snapshot where an individual snapshot is taken per VM and stored with the VM.
- If I want to recover a VM from this LUN level snapshot whats the process? Can I restore an individual VM from the LUN Level snapshot back to its original LUN where other VMs are probably still running, without affecting them?
- Is it possible I could lose my base LUN ever, maybe corruption, accidental delete, malicious delete? If so, does that render my snapshots useless as its taken from the base?
- Does my snapshot LUN need to be atleast the same size as the LUN I want to snapshot
- Can you point me to some easy reading/videos/training material that might show me more about what I am trying learn?
if this thread progresses well I hope to be able to start asking more challenging questions around this subject... such as introducing a second 3par in a DR site etc...
I wished I'd payed more attention in my training now... grrrr... idiot!
Understanding Snapshots
Re: Understanding Snapshots
(Note: I use the term "volume" instead of "LUN", since there is a slight change in parlance with 3PAR, and this reduces the overloading of terminology.)
Yes, the Copy CPG is just the specified CPG where snapshots of that base volume will be stored. You can make it the same CPG as the base volume's User CPG, but many (including myself) have dedicated CPGs for the purpose of snapshots. I do this to use a different class of disk. You can change the Copy CPG after the fact if you tunevv that volume (and you have the Dynamic Optimization license).
You would need to resignature the volume to mount the snapshot's VMFS, if mounting it into the same datacenter. This is because the snapshot would be out of sync with the metadata contained in that VMware datacenter about that particular volume. There's some brief notes about it from VMware here: http://kb.vmware.com/selfservice/micros ... Id=1011387
In the event of accidental or malicious delete, the snapshot volumes are removed along with the base volume. In the event of corruption, your snapshots will be similarly corrupted (assuming that it has affected an area where a R/W snapshot has not yet overwritten).
You have no choice about the size of your snapshot—it's the same size as the base volume at the time the snapshot was taken. You cannot resize snapshots, even if you resize the base volume.
When using a 3par snapshot, I'm assuming the whole LUN is snapshot to the COPY CPG? As opposed to a VMware snapshot where an individual snapshot is taken per VM and stored with the VM.
Yes, the Copy CPG is just the specified CPG where snapshots of that base volume will be stored. You can make it the same CPG as the base volume's User CPG, but many (including myself) have dedicated CPGs for the purpose of snapshots. I do this to use a different class of disk. You can change the Copy CPG after the fact if you tunevv that volume (and you have the Dynamic Optimization license).
If I want to recover a VM from this LUN level snapshot what's the process? Can I restore an individual VM from the LUN Level snapshot back to its original LUN where other VMs are probably still running, without affecting them?
You would need to resignature the volume to mount the snapshot's VMFS, if mounting it into the same datacenter. This is because the snapshot would be out of sync with the metadata contained in that VMware datacenter about that particular volume. There's some brief notes about it from VMware here: http://kb.vmware.com/selfservice/micros ... Id=1011387
Is it possible I could lose my base LUN ever, maybe corruption, accidental delete, malicious delete? If so, does that render my snapshots useless as its taken from the base?
In the event of accidental or malicious delete, the snapshot volumes are removed along with the base volume. In the event of corruption, your snapshots will be similarly corrupted (assuming that it has affected an area where a R/W snapshot has not yet overwritten).
Does my snapshot LUN need to be at least the same size as the LUN I want to snapshot
You have no choice about the size of your snapshot—it's the same size as the base volume at the time the snapshot was taken. You cannot resize snapshots, even if you resize the base volume.
- Richard Siemers
- Site Admin
- Posts: 1333
- Joined: Tue Aug 18, 2009 10:35 pm
- Location: Dallas, Texas
Re: Understanding Snapshots
Rooey wrote:When using a 3par snapshot, I'm assuming the whole LUN is snapshot to the COPY CPG? As opposed to a VMware snapshot where an individual snapshot is taken per VM and stored with the VM.
The "whole" lun is not copied to the CopyCPG, just the deltas or changes made since the snapshot was created. If the original volume is pretty static, with little to no changes... the cost in megabytes of the snapshot will be quite small, if its a drive with a high change rate, such as an oracle redo log, then the snapshot will probably take up as much copycpg space as the source volume uses within a few hours. The snapshot uses a "copy on first write" method to preserve the data. When a snapshot is created, it starts basically as copy of the original volume's table of contents, with read only pointers back to the original blocks of data. If the original volume needs to edit a data that has a snapshot pointer back to it... it copies that block down to the CopyCPG and then lets the original volume change that block as much as it wants to. If a host has a read/write snapshot mounted up and wants to make a change to a block that is a pointer to original volume, then the pointer is dropped and replaced with a real block allocated into the CopyCPG.
If I want to recover a VM from this LUN level snapshot whats the process? Can I restore an individual VM from the LUN Level snapshot back to its original LUN where other VMs are probably still running, without affecting them?
In addition to what was already said, HP makes a special application called Recovery Manager for VMware that will manage and automate most of your VMware related snaps and recoveries. I've never used it, but I am sure someone on these forums could shares some opinions on whether or not its a solid solution or suggest alternatives.
Is it possible I could lose my base LUN ever, maybe corruption, accidental delete, malicious delete? If so, does that render my snapshots useless as its taken from the base?
Depends on what the corruption is... if you use Raid0 cpgs and your lose a drive... or by some malicious admin, lose your CPG or yank several magazines of drives from the array... snapshots are not a DR solution unless you are also using Remote Copy (which basically takes a snap and replicates the snap to another 3PAR). If your host somehow manages to corrupt the VMFS datastore, or SQL datafile, etc... "logical corruption" then yes a snapshot would be an ideal method of quickly restoring.
Does my snapshot LUN need to be atleast the same size as the LUN I want to snapshot
As previously stated, you can't edit the size of snapshot... it inherits it size from the source... HOWEVER, if you are using snapshot copies for building duplicate instances, such as a snapshot of a production database mounted up on a dev host... you CAN promote a snapshot to regular volume, then grow the LUN as needed. Cloning is also an option, it uses a snapshot to copy a volume at an exact point in time, then copies the snapshot into a stand alone volume. The clone is immediately available for use even before the copy is finished due to the snapshot.
I don't know about "easy", but the 3PAR Storserv Concepts Guide is a good start to a 3PAR overview of all its features.Can you point me to some easy reading/videos/training material that might show me more about what I am trying learn?
http://h20566.www2.hpe.com/hpsc/doc/pub ... n-us&cc=us
Richard Siemers
The views and opinions expressed are my own and do not necessarily reflect those of my employer.
The views and opinions expressed are my own and do not necessarily reflect those of my employer.
Re: Understanding Snapshots
Chaps
Thank you. Thank you very much. I need to read and reread the info you have given me and try and absorb this to be able to ask further questions for me to start understanding this fully. I also do have Remote Copy but don't want to introduce that into the thread yet as I think I'll get lost! I want snapshots to supplement a potential backup/recovery design where I can restore VM's reasonably quickly from scheduled 6 hour snapshots if required in the first instance... and failing this go to our BUR solution which will only allow full restore from 24 hours ago. ..
Hope I'm making sense! And you aren't nodding your heads thinking I'm mad!
Thank you. Thank you very much. I need to read and reread the info you have given me and try and absorb this to be able to ask further questions for me to start understanding this fully. I also do have Remote Copy but don't want to introduce that into the thread yet as I think I'll get lost! I want snapshots to supplement a potential backup/recovery design where I can restore VM's reasonably quickly from scheduled 6 hour snapshots if required in the first instance... and failing this go to our BUR solution which will only allow full restore from 24 hours ago. ..
Hope I'm making sense! And you aren't nodding your heads thinking I'm mad!
Re: Understanding Snapshots
HP with Recovery Manager Central for VMware provides this functionality from vCentre.
http://www8.hp.com/h20195/V2/getpdf.aspx/4AA5-5864ENW.pdf?ver=1.0
(You can ignore the Express Protect / Flat Backup feature and just use snaps if you don't have StoreOnce)
Also Veeam supports a similar feature.
http://www.veeam.com/san-snapshots-explorer.html
http://www8.hp.com/h20195/V2/getpdf.aspx/4AA5-5864ENW.pdf?ver=1.0
(You can ignore the Express Protect / Flat Backup feature and just use snaps if you don't have StoreOnce)
Also Veeam supports a similar feature.
http://www.veeam.com/san-snapshots-explorer.html
Re: Understanding Snapshots
Yep, I understand HP RMC can do this... everyone keeps telling me this and I will be getting it...
But i really have this need (OCD) to be able to somehow just take snapshots/virtual copies on a schedule and test recovery of a VM with what I have right now...
I am now thinking it might be better if I schedule VMware snapshots to snapshot each virtual machine (as I can get an Application Consistent Snapshot this way) and then Virtual Copy the underlying LUN. If the VM was ever corrupted and unrecoverable, then I think I could present and mount the Virtual Copy LUN as a new datastore to my ESXi Cluster, browse to the VM (which is a snapshot from before corruption) and add it into my vCenter inventory (after deleting the corrupted version of the VM from vCenter)... I would hope I can then storage vMotion the VM back to its orginal High Performance LUN and unmount the Virtual Copy LUN, as its purpose was to only recover the corrupted VM...
Someone give me a gold star if this makes sense ... I understand it might be convoluted... but I'm going for understanding and learning right now seeing as I have been given ownership of a couple of 3Pars, so I might as well learn them while I have the opportunity!... RMC for me is about 4 months away...
Automating this will cause me further challenges and reaching out to bug you guys no doubt! But if someone can confirm the approach makes sense, then atleast I am understanding what I am reading
But i really have this need (OCD) to be able to somehow just take snapshots/virtual copies on a schedule and test recovery of a VM with what I have right now...
I am now thinking it might be better if I schedule VMware snapshots to snapshot each virtual machine (as I can get an Application Consistent Snapshot this way) and then Virtual Copy the underlying LUN. If the VM was ever corrupted and unrecoverable, then I think I could present and mount the Virtual Copy LUN as a new datastore to my ESXi Cluster, browse to the VM (which is a snapshot from before corruption) and add it into my vCenter inventory (after deleting the corrupted version of the VM from vCenter)... I would hope I can then storage vMotion the VM back to its orginal High Performance LUN and unmount the Virtual Copy LUN, as its purpose was to only recover the corrupted VM...
Someone give me a gold star if this makes sense ... I understand it might be convoluted... but I'm going for understanding and learning right now seeing as I have been given ownership of a couple of 3Pars, so I might as well learn them while I have the opportunity!... RMC for me is about 4 months away...
Automating this will cause me further challenges and reaching out to bug you guys no doubt! But if someone can confirm the approach makes sense, then atleast I am understanding what I am reading
Re: Understanding Snapshots
Just came across this great post as I have stumbled across a bunch of snapshots taken a year ago for our environment.
The Creation Time stamp, is the time it was made, is there a way to verify they are actively being used still?
Could I delete just these snapshots under "Restore Points" in 3Par StoreServ and create new ones?
The Creation Time stamp, is the time it was made, is there a way to verify they are actively being used still?
Could I delete just these snapshots under "Restore Points" in 3Par StoreServ and create new ones?
Re: Understanding Snapshots
Does the VC (snapshot) get assigned a UDID?
Re: Understanding Snapshots
When I create a new vv, and assign a snapshot CPG to it, the vv appears in both the normal user CPG AND the snapshot CPG, without any snaps being present. It is quite frankly bizarre.
As soon as I take a snap, the lun which appears in the Snapshot CPG changes to the name of the snap, and again, on deletion of the snap, the normal vv name appears.
Is this normal to have a volume registered over several CPGs like this, even though theoretically its not actually using it since its not been snapshotted. Its like its just "registered" with that CPG for future reference, but its a really odd way of doing it and not intuitive to me at least. Am I going mad?
As soon as I take a snap, the lun which appears in the Snapshot CPG changes to the name of the snap, and again, on deletion of the snap, the normal vv name appears.
Is this normal to have a volume registered over several CPGs like this, even though theoretically its not actually using it since its not been snapshotted. Its like its just "registered" with that CPG for future reference, but its a really odd way of doing it and not intuitive to me at least. Am I going mad?
Re: Understanding Snapshots
T16 wrote:When I create a new vv, and assign a snapshot CPG to it, the vv appears in both the normal user CPG AND the snapshot CPG, without any snaps being present. It is quite frankly bizarre.
As soon as I take a snap, the lun which appears in the Snapshot CPG changes to the name of the snap, and again, on deletion of the snap, the normal vv name appears.
Is this normal to have a volume registered over several CPGs like this, even though theoretically its not actually using it since its not been snapshotted. Its like its just "registered" with that CPG for future reference, but its a really odd way of doing it and not intuitive to me at least. Am I going mad?
To be honest you seem to be doing a lot but having really gotten the concept of 3PAR. May I ask what sort of storage arrays you are used to managing and I would generally recommend reading the 3PAR Storage Concept Guide. It might help you get a better view of what you are managing.
The views and opinions expressed are my own and do not necessarily reflect those of my current or previous employers.