[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [PROPOSE] Backup and Recovery Framework


First of all, the framework is agnostic of any backup vendor/provider. It just provides the way to communicate with the provider to allow users to backup their VMs.


Will is right, backups are not the same as snapshots, I got confused with the Veeam backup which starts from taking a snapshot to generate the backup. The way they are generated are up to each vendor, and it is not relevant to the framework. Sorry for the confusion, we are also introducing the Veeam plugin in top of this framework.


So, in terms of this framework, it does not matter what a backup is, we just keep a reference to an existing backup on each vendor.


I would also like to state that this framework is not using a snapshot mechanism for backups in CloudStack. It is just providing interfaces and mechanisms for vendors/providers plugins in CloudStack. Implementation of each plugin should be relevant to the vendor and any backup handling will be external to CloudStack.


I think that the diagram in the FS could explain what the task of the framework is: https://cwiki.apache.org/confluence/download/attachments/85476500/Backup%20and%20Recovery%20UML.png?version=1&modificationDate=1528367734183&api=v2. The framework simply allows management server to communicate with backup and recovery vendors/providers. To allow CloudStack communication with an specific vendor, a plugin should be implemented for it (as we are doing with Veeam for example).


Please let me know if it is still confusing and I'll try to explain better

________________________________
From: Todd Pigram <todd@xxxxxxxxxxxxxx>
Sent: Thursday, June 7, 2018 12:11:18 PM
To: dev@xxxxxxxxxxxxxxxxxxxxx
Subject: Re: [PROPOSE] Backup and Recovery Framework

Now outside of our ACS, I use Veeam for vCloud and vSphere and use Alike
for our XS pools. Inside ACS, the VMs use MaxBackup


Nicolas.Vazquez@xxxxxxxxxxxxx 
www.shapeblue.com
,   
@shapeblue
  
 

On Thu, Jun 7, 2018 at 11:09 AM, Todd Pigram <todd@xxxxxxxxxxxxxx> wrote:

> I agree completely. That is why I use inside VM backup products currently.
>
> On Thu, Jun 7, 2018 at 11:07 AM, Will Stevens <wstevens@xxxxxxxxxxxx>
> wrote:
>
>> I have not been following this thread entirely, but I would not count on
>> our snapshot mechanism for backups.  If something like Veeam is used, then
>> an agent is in play and that agent will quiesce the running processes,
>> such
>> as MySQL.  This is important in order to do backup and DR as you will lose
>> transactions in memory and can potentially have a corrupted backup.
>>
>> Snapshots are not backups!!!
>>
>> *Will Stevens*
>> Chief Technology Officer
>> c 514.826.0190
>>
>> <https://goo.gl/NYZ8KK>
>>
>>
>> On Thu, Jun 7, 2018 at 10:13 AM Rafael Weingärtner <
>> rafaelweingartner@xxxxxxxxx> wrote:
>>
>> > Ok, now I start to understand.  So, we will be handling snapshots and
>> they
>> > will be sent to a third-party system.
>> >
>> > Now let’s proceed. If I configure this “feature” with external=false.
>> What
>> > does it do? Nothing? I mean, what is the purpose of this “external”
>> > parameter?
>> >
>> >
>> > On Thu, Jun 7, 2018 at 4:09 PM, Nicolas Vazquez <
>> > Nicolas.Vazquez@xxxxxxxxxxxxx> wrote:
>> >
>> > > Its format may depend on the provider, but under the hood it is a
>> > snapshot
>> > > which can be full or incremental and it is copied and stored in a
>> backup
>> > > repository in one of the backup servers. For example, if you use
>> Veeam,
>> > you
>> > > will find your backups stored as .vbk (for full backups) and .vib (for
>> > > incremental backups). It is the task of the backup provider to create
>> and
>> > > move these files to the backup repository.
>> > >
>> > > ________________________________
>> > > From: Rafael Weingärtner <rafaelweingartner@xxxxxxxxx>
>> > > Sent: Thursday, June 7, 2018 10:39:35 AM
>> > > To: dev
>> > > Subject: Re: [PROPOSE] Backup and Recovery Framework
>> > >
>> > > Ok, there are a lot of things, which are kind of blurry to me. So,
>> let’s
>> > > try to go step by step.
>> > >
>> > > Let me ask again. What is a backup of the VM? You mention
>> restore-point.
>> > > What is it? Is it a snapshot in the hypervisor (complete/partial
>> > snapshot)?
>> > > If it is not a snapshot,  what is it then? Are you installing some
>> agent
>> > in
>> > > the VM and then copying/mirroring the files in the VM’s file system?
>> > >
>> > > On Thu, Jun 7, 2018 at 3:26 PM, Nicolas Vazquez <
>> > > Nicolas.Vazquez@xxxxxxxxxxxxx> wrote:
>> > >
>> > > > Hi Daniel and Rafael,
>> > > >
>> > > >
>> > > > I have updated the diagram picture on the FS, please let me know if
>> the
>> > > > problem persists.
>> > > >
>> > > >
>> > > > @<mailto:daniel.herrmann@xxxxxxxxxxxxxxxx>Daniel you are right
>> about
>> > > > restoring VMs, please let me update the FS. Regarding your
>> question, we
>> > > > have considered the backups as a first class citizens, meaning that
>> a
>> > > > backup would continue existing unless its originating VM has been
>> > > > removed/expunged. This way, listBackups API method would list them
>> in
>> > > order
>> > > > to allow a user to restore a deleted VM from a backup.
>> > > >
>> > > >
>> > > > @Rafael the idea of the Dummy provider is simply returning the
>> expected
>> > > > entities from a backup provider. For example, the BackupManager
>> could
>> > ask
>> > > > the BackupProvider to list its backups and would expect a list of
>> > > backups.
>> > > > As the Dummy Provider implements the BackupProvider interface, it
>> has
>> > its
>> > > > own implementation of listBackups() method and simply returns some
>> mock
>> > > > entities to the Backup Manager. It does not store any backup, it
>> simply
>> > > > creates them in execution time and returns them.
>> > > >
>> > > >
>> > > > From the CloudStack point of view, Backups are stored externally, on
>> > the
>> > > > Backup and Recovery provider servers and are handled by them. We
>> keep
>> > > track
>> > > > of the existing backups in the Backup provider. Implementation may
>> > depend
>> > > > on the Backup and Recovery provider, would consist on snapshots, but
>> > are
>> > > > handled by the provider and not by CloudStack.
>> > > >
>> > > >
>> > > > Backup policies are created on the Backup and Recovery provider and
>> > > > provide retention options and periods as well as backup times. For
>> > > example,
>> > > > the provider may expose Golden, Silver and Bronze policies and have
>> > > > different options for each of them.
>> > > >
>> > > >
>> > > > The 'external' parameter on the listBackupPolicies API method would
>> > only
>> > > > ask the Backup and Recovery provider to list their policies if it is
>> > set
>> > > to
>> > > > true. If it is not set, or set to false, it just simply returns the
>> > > > existing mapped policies in CloudStack. Initially, CloudStack does
>> not
>> > > > store any backup policy, those should be imported/mapped by the
>> > > > 'importBackupPolicy' API method.
>> > > >
>> > > >
>> > > > Finally, we consider backups as restore points. As mentioned above,
>> > > > backups are stored on the Backup and Recovery provider servers, so
>> we
>> > > keep
>> > > > a mapping between the CloudStack entity and the backup, which can be
>> > used
>> > > > to restore VMs.
>> > > >
>> > > >
>> > > > Please let me know if anything.
>> > > >
>> > > >
>> > > > Regards,
>> > > >
>> > > > Nicolas Vazquez
>> > > >
>> > > > ________________________________
>> > > > From: Rafael Weingärtner <rafaelweingartner@xxxxxxxxx>
>> > > > Sent: Thursday, June 7, 2018 6:55:57 AM
>> > > > To: dev
>> > > > Subject: Re: [PROPOSE] Backup and Recovery Framework
>> > > >
>> > > > Hi Nicolas,
>> > > >
>> > > > The idea is great. I have some questions though.
>> > > >
>> > > > Before getting to the questions, your pictures are broken, or at
>> least
>> > > they
>> > > > seem broken to me.
>> > > >
>> > > > How would the “dummy” backup provider work? I mean, what does it do?
>> > > where
>> > > > does it storage the "backup"?
>> > > > What is the difference from this Backup to a Snapshot? Is it simply
>> > > storing
>> > > > a snapshot outside the cloud structure?
>> > > > How would these “backup policies” look like? I mean, what can I
>> > > configure?
>> > > > Regarding that parameter “external” of your backup policy, how
>> would it
>> > > > work if we set it to false?
>> > > > What are you defining as a backup (what do you mean by backup in the
>> > > > context of this feature?)? I mean, is it simply a snapshot file that
>> > you
>> > > > store somewhere else?
>> > > >
>> > > >
>> > > > Nicolas.Vazquez@xxxxxxxxxxxxx
>> > > > www.shapeblue.com<http://www.shapeblue.com>
>> > > > ,
>> > > > @shapeblue
>> > > >
>> > > >
>> > > >
>> > >
>> > > Nicolas.Vazquez@xxxxxxxxxxxxx
>> > > www.shapeblue.com<http://www.shapeblue.com>
>> > > ,
>> > > @shapeblue
>> > >
>> > >
>> > >
>> > > > On Thu, Jun 7, 2018 at 6:05 AM, <daniel.herrmann@xxxxxxxxxxxxxxxx>
>> > > wrote:
>> > > >
>> > > > > Hi all,
>> > > > >
>> > > > > +1 for this feature request, thank you for bringing this to life!
>> > > Volume
>> > > > > snapshots are what our users currently use as backup solution,
>> which
>> > > has
>> > > > > some inherent problems such as handling large volumes and the
>> > handling
>> > > of
>> > > > > multiple volumes (consistency).
>> > > > >
>> > > > > One thought: when restoring a VM, I think there are two cases to
>> > > > consider:
>> > > > >
>> > > > > - The VM is still in CS and the user just wants to restore an old
>> > > version
>> > > > > of the VM
>> > > > > - The VM has already been deleted and the user wants to restore
>> the
>> > > > > deleted VM
>> > > > >
>> > > > > As listBackups is described as " listBackups API method. List
>> > existing
>> > > > > backups for a VM". Is the second use case supported at all?
>> > > > >
>> > > > > I like the approach not to define the backup policies in CS but in
>> > > > > whatever tool is used in the end, as otherwise a lot of features
>> > might
>> > > be
>> > > > > lost. We as services providers need to create them in the first
>> > place,
>> > > > but
>> > > > > I think this is the better solution.
>> > > > >
>> > > > > Regards
>> > > > > Daniel
>> > > > >
>> > > > > On 06.06.18, 14:43, "Nicolas Vazquez" <
>> > Nicolas.Vazquez@xxxxxxxxxxxxx>
>> > > > > wrote:
>> > > > >
>> > > > >     Hi all,
>> > > > >
>> > > > >
>> > > > >     We would like to introduce a new framework into CloudStack
>> that
>> > > will
>> > > > > allow adding Backup and Recovery providers as plugins. This
>> framework
>> > > > goal
>> > > > > is allowing users to backup their guest VMs for recovery purposes.
>> > > > >
>> > > > >
>> > > > >     Please find the FS on this link: https://cwiki.apache.org/
>> > > > > confluence/display/CLOUDSTACK/Backup+and+Recovery+Framework
>> > > > >
>> > > > >
>> > > > >     Regards,
>> > > > >
>> > > > >     Nicolas Vazquez
>> > > > >
>> > > > >     Nicolas.Vazquez@xxxxxxxxxxxxx
>> > > > >     www.shapeblue.com<http://www.shapeblue.com>
>> > > > >     ,
>> > > > >     @shapeblue
>> > > > >
>> > > > >
>> > > > >
>> > > > >
>> > > > >
>> > > > >
>> > > >
>> > > >
>> > > > --
>> > > > Rafael Weingärtner
>> > > >
>> > >
>> > >
>> > >
>> > > --
>> > > Rafael Weingärtner
>> > >
>> >
>> >
>> >
>> > --
>> > Rafael Weingärtner
>> >
>>
>
>
>
> --
>
>
> Todd Pigram
> http://about.me/ToddPigram
> www.linkedin.com/in/toddpigram/<http://www.linkedin.com/in/toddpigram/>
> @pigram86 on twitter
> https://plus.google.com/+ToddPigram86
> Mobile - 216-224-5769
>
> PGP Public Key
> <http://pgp.mit.edu/pks/lookup?op=get&search=0x96B7B0F0C55933BB>
>
>


--


Todd Pigram
http://about.me/ToddPigram
www.linkedin.com/in/toddpigram/<http://www.linkedin.com/in/toddpigram/>
@pigram86 on twitter
https://plus.google.com/+ToddPigram86
Mobile - 216-224-5769

PGP Public Key
<http://pgp.mit.edu/pks/lookup?op=get&search=0x96B7B0F0C55933BB>