AWS Cloud Protection Manager Part 3: Backup and Restore

Backup

Backups are created according to the schedule specified in the backup policy. We discussed how to configure backup policies in the previous blog post of the series. The list of backups you see on the Backup Monitor tab are your restore points. Backups that are older then the specified retention policy will be purged from the list and you will not see them there, unless you move them to “Freezer”.

It is important to understand that apart from volume snapshots, for each backed up instance CPM also creates an AMI. Those who has hands-on experience with AWS may already know, that AMIs is the only way to create clones of Windows EC2 instances in AWS. If you go to AWS console and try to find a clone action under the instance Action menu, you won’t find any. You will have “Create Image” instead. It creates an AMI, from which you can then spin up a clone of an instance the image was created from.

CPM does exactly that. For each backup policy the instance is under, it creates one AMI. In our example we have four backup policies, that will result in four AMIs for each of the instances. Every AMI has to have at least one storage volume. So CPM will include the root volume of each instance into AMI, just because it has to. But AMIs are required only to restore EC2 instance configuration. Data is restored from volume snapshots, that can be used to create new volumes from them and then attach them to the instance. You can click on the View button under Snapshots to find the corresponding snapshot and AMI IDs.

There is a backup log for each job run as well that is helpful for issue troubleshooting.

Restore

To perform a restore click on the Recover button next to the backup job and you will get the list of the instances you can recover. CPM offers you three options: instance recovery, volume recovery and file recovery. Let’s go back to front.

File recovery is probably the most used recovery option. As it lets you restore individual files. When you click on the “Explore” button, CPM creates new volumes from the snapshots you are restoring from and mount them to the CPM instance. You are then presented with a simple file system browser where you can find the file and click on the green down arrow icon in Download column to save the file to your computer.

If you click on “Volume Only”, you can restore particular volumes. Restored volumes are not attached to any instance, unless you specify it under “Attach to Instance” column. You can then select under “Attach Behaviour” what CPM should do if such volume is already attached to the instance or if you want to automatically detach the original volume, but the instance is running (you can do it only if instance is stopped).

And the last option is “Instance”. It will create a clone of the original instance using the pre-generated AMI and volume snapshots, as we discussed in the Backup section of this blog post. You can specify many options under Advanced Options section, including recovery to another VPC or different availability zone. If anything, make sure you specify a new IP address for the instance, otherwise you’ll have a conflict and your restore will fail. Ideally you should also shut down the original EC2 instance before spinning up a restore clone.

Advanced Features

There are quite a few worth mentioning. So far we have looked at simple EC2 instance restore. But you don’t have to backup whole instances, you can also backup individual volumes. On top of that, CPM supports RDS database, Aurora and Redshift cluster backups.

If you run MS Exchange, Sharepoint or SQL on your EC2 instances, you can install CPM backup agent on them to ensure you have application-consistent backups via VSS, as opposed to crash-consistent backups you get if agent is not used. If you install the agent, you can also run a script on the instance before and after the backup is taken.

Last but not least is DR. Restoring to another availability zone within the region is already supported on instance recovery level. You can choose availability zone you want to restore to. It is not possible to recover to another region, though. Because AWS snapshots and AMIs are local to the region they are created in. If you want to be able to recover to another region, you can configure DR in CPM, which will utilise AWS AMI and snapshot copy functionality to copy backups to another region at configured frequency.

Conclusion

Overall, I found Cloud Protection Manager very easy to install, configure and use. If you come from infrastructure background, at first glance CPM may look to you like a very basic tool, compared to such feature-rich solutions like Veeam or Commvault. But that feeling is misleading. CPM is simple, because AWS simple. All infrastructure complexity is hidden under the covers. As a result, all AWS backup tools need to do is create snapshots and CPM does it well.

Advertisement

AWS Cloud Protection Manager Part 1: Intro to Snapshots

Overview

We have all been dealing with on-premises infrastructure for years before public cloud became a thing. Depending on complexity and size of your environment, retention and other requirements you can choose from plethora of products, such as Veeam, Commvault, NetWorker, you name it.

When public cloud started getting more traction we realised that the paradigm has shifted. Public cloud providers are pushing us to treat instances of application servers as cattle versus pets and build fault tolerance into code and not rely on underlying infrastructure reliability.

If application is no longer monolithic and is distributed among multiple instances of the same server, loss of any one instance has little overall impact on the application and does not require a restore. We can simply spin up a new instance instead.

However, majority of applications in today’s IT environments are still monolithic and need to be backed up. Data needs to be protected as well, if it gets corrupted or if for any other reason we need to roll back our application to a certain point in time, backups are necessary.

AWS Snapshots

When I started researching the topic of backup in the cloud I realised that it is still a wild west. Speaking of AWS specifically, if you are using one of the native AWS services, such as RDS, you would usually have the backup feature built-in. Otherwise, if you simply want to backup some application or service running inside an AWS instance, all you have is snapshots.

We have all heard the mantra, that snapshot is not a backup, because usually it is kept in the same place as the original data it was created from. This is true for most of the traditional storage arrays. AWS is slightly different.

Typically when you start an EC2 instance, its disk volumes are kept on Elastic Block Store (EBS), which is an equivalent of block storage in traditional infrastructure world. But when you create a snapshot of an EBS volume, the snapshot is transferred to S3 object storage.

When the first snapshot is created, the whole volume needs to be copied to S3. It may take some time to complete, depending on the volume size, but the copy process happens in the background and doesn’t impact the original EC2 instance. All subsequent snapshots will be significantly quicker, unless you overwrite large amount of data between snapshots.

CloudWatch Event Rules

As we can see, unlike traditional storage arrays, AWS snapshots are kept separately from the original data. They can even be restored to another availability zone or copied to another region to protect from region-wide failures, which makes AWS snapshots a decent backup solution. But devil is in the details.

Creating a volume snapshot manually is simple, you pick “Create Snapshot” from the Actions menu and the rest is handled by AWS. To automate the process, AWS offers CloudWatch event rules for scheduled snapshot creation. By going to CloudWatch > Events > Create Rule, setting up a schedule, selecting “EC2 CreateSnapshot API call” and the volume ID in the Target section you can implement a simple backup in AWS.

However, there are two immediate problems with such approach. The first is scalability of it. Snapshots in AWS are created on volumes, not instances. This has interesting implications, such as it is not possible to create a consistent snapshot across multiple volumes. But more importantly, if you have more than one volume on every instance, you may end up having dozens of CloudWatch rules, that are hard to manage and keep track of.

Second is retention. CloudWatch events let you create snapshots, but you cannot specify how long you want to keep them for. You will need to write your own scripts using AWS APIs to delete snapshots, which makes the whole solution questionable, as you can then use APIs to create snapshots as well.

Conclusion

As you can see, AWS does not offer a simple out of the box solution for EC2 backup. You either need to write your own scripts using AWS APIs or lean on backup solutions built specifically to solve the problem of backups in AWS. One of such solutions Cloud Protection Manager we will discuss in the next blog post of the series.

NetApp NVRAM and Write Caching

Overview

NetApp storage systems use several types of memory for data caching. Non-volatile battery-backed memory (NVRAM) is used for write caching (whereas main memory and flash memory in forms of either extension PCIe card or SSD drives is used for read caching). Before going to hard drives all writes are cached in NVRAM. NVRAM memory is split in half and each time 50% of NVRAM gets full, writes are being cached to the second half, while the first half is being written to disks. If during 10 seconds interval NVRAM doesn’t get full, it is forced to flush by a system timer.

To be more precise, when data block comes into NetApp it’s actually written to main memory and then journaled in NVRAM. NVRAM here serves as a backup, in case filer fails. When data has been written to disks as part of so called Consistency Point (CP), write blocks which were cached in main memory become the first target to be evicted and replaced by other data.

Caching Approach

NetApp is frequently criticized for small amounts of write cache. For example FAS3140 has only 512MB of NVRAM, FAS3220 has a bit more 1,6GB. In mirrored HA or MetroCluster configurations NVRAM is mirrored via NVRAM interconnect adapter. Half of the NVRAM is used for local operations and another half for the partner’s. In this case the amount of write cache becomes even smaller. In FAS32xx series NVRAM has been integrated into main memory and is now called NVMEM. You can check the amount of NVRAM/NVMEM in your filer by running:

> sysconfig -a

The are two answers to the question why NetApp includes less cache in their controllers. The first one is given in white paper called “Optimizing Storage Performance and Cost with Intelligent Caching“. It states that NetApp uses different approach to write caching, compared to other vendors. Most often when data block comes in, cache is used to keep the 8KB data block, as well as 8KB inode and 8KB indirect block for large files. This way, write cache can be thought as part of the physical file system, because it mimics its structure. NetApp on the other hand uses journaling approach. When data block is received by the filer, 8KB data block is cached along with 120B header. Header contains all the information needed to replay the operation. After each cache flush Consistency Point (CP) is created, which is a special type of consistent file system snapshot. If controller fails, the only thing which needs to be done is reverting file system to the latest consistency point and replaying the log.

But this white paper was written in 2010. And cache journaling is not a feature unique to NetApp. Many vendors are now using it. The other answer, which makes more sense, was found on one of the toaster mailing list archives here: NVRAM weirdness (UNCLASSIFIED). I’ll just quote the answer:

The reason it’s so small compared to most arrays is because of WAFL. We don’t need that much NVRAM because when writes happen, ONTAP writes out single complete RAID stripes and calculates parity in memory. If there was a need to do lots of reads to regenerate parity, then we’d have to increase the NVRAM more to smooth out performance.

NVLOG Shipping

A feature called NVLOG shipping is an integral part of sync and semi-sync SnapMirror. NVLOG shipping is simply a transfer of NVRAM writes from the primary to a secondary storage system.  Writes on primary cannot be transferred directly to NVRAM of the secondary system, because in contrast to mirrored HA and MetroCluster, SnapMirror doesn’t have any hardware implementation of the NVRAM mirroring. That’s why the stream of data is firstly written to the special files on the volume’s parent aggregate on the secondary system and then are read to the NVRAM.

Documents I found useful:

WP-7107: Optimizing Storage Performance and Cost with Intelligent Caching

TR-3326: 7-Mode SnapMirror Sync and SnapMirror Semi-Sync Overview and Design Considerations

TR-3548: Best Practices for MetroCluster Design and Implementation

United States Patent 7730153: Efficient use of NVRAM during takeover in a node cluster

Magic behind NetApp VSC Backup/Restore

NetApp Virtual Storage Console is a plug-in for VMware vCenter which provides capabilities to perform instant backup/restore using NetApp snapshots. It uses several underlying NetApp features to accomplish its tasks, which I want to describe here.

Backup Process

When you configure a backup job in VSC, what VSC does, is it simply creates a NetApp snapshot for a target volume on a NetApp filer. Interestingly, if you have two VMFS datastores inside one volume, then both LUNs will be snapshotted, since snapshots are done on the volume level. But during the datastore restore, the second volume will be left intact. You would think that if VSC reverts the volume to the previously made snapshot, then both datastores should be affected, but that’s not the case, because VSC uses Single File SnapRestore to restore the LUN (this will be explained below). Creating several VMFS LUNs inside one volume is not a best practice. But it’s good to know that VSC works correctly in this case.

Same thing for VMs. There is no sense of backing up one VM in a datastore, because VSC will make a volume snapshot anyway. Backup the whole datastore in that case.

Datastore Restore

After a backup is done, you have three restore options. The first and least useful kind is a datastore restore. The only use case for such restore that I can think of is disaster recovery. But usually disaster recovery procedures are separate from backups and are based on replication to a disaster recovery site.

VSC uses NetApp’s Single File SnapRestore (SFSR) feature to restore a datastore. In case of a SAN implementation, SFSR reverts only the required LUN from snapshot to its previous state instead of the whole volume. My guess is that SnapRestore uses LUN clone/split functionality in background, to create new LUN from the snapshot, then swap the old with the new and then delete the old. But I haven’t found a clear answer to that question.

For that functionality to work, you need a SnapRestore license. In fact, you can do the same trick manually by issuing a SnapRestore command:

> snap restore -t file -s nightly.0 /vol/vol_name/vmfs_lun_name

If you have only one LUN in the volume (and you have to), then you can simply restore the whole volume with the same effect:

> snap restore -t vol -s nightly.0 /vol/vol_name

VM Restore

VM restore is also a bit controversial way of restoring data. Because it completely removes the old VM. There is no way to keep the old .vmdks. You can use another datastore for particular virtual hard drives to restore, but it doesn’t keep the old .vmdks even in this case.

VSC uses another mechanism to perform VM restore. It creates a LUN clone (don’t confuse with FlexClone,which is a volume cloning feature) from a snapshot. LUN clone doesn’t use any additional space on the filer, because its data is mapped to the blocks which sit inside the snapshot. Then VSC maps the new LUN to the ESXi host, which you specify in the restore job wizard. When datastore is accessible to the ESXi host, VSC simply removes the old VMDKs and performs a storage vMotion from the clone to the active datastore (or the one you specify in the job). Then clone is removed as part of a clean up process.

The equivalent cli command for that is:

> lun clone create /vol/clone_vol_name -o noreserve -b /vol/vol_name nightly.0

Backup Mount

Probably the most useful way of recovery. VSC allows you to mount the backup to a particular ESXi host and do whatever you want with the .vmdks. After the mount you can connect a virtual disk to the same or another virtual machine and recover the data you need.

If you want to connect the disk to the original VM, make sure you changed the disk UUID, otherwise VM won’t boot. Connect to the ESXi console and run:

# vmkfstools -J setuuid /vmfs/volumes/datastore/VM/vm.vmdk

Backup mount uses the same LUN cloning feature. LUN is cloned from a snapshot and is connected as a datastore. After an unmount LUN clone is destroyed.

Some Notes

VSC doesn’t do a good cleanup after a restore. As part of the LUN mapping to the ESXi hosts, VSC creates new igroups on the NetApp filer, which it doesn’t delete after the restore is completed.

What’s more interesting, when you restore a VM, VSC deletes .vmdks of the old VM, but leaves all the other files: .vmx, .log, .nvram, etc. in place. Instead of completely substituting VM’s folder, it creates a new folder vmname_1 and copies everything into it. So if you use VSC now and then, you will have these old folders left behind.

Mounting VMware Virtual Disks

There are millions of posts on that topic all over the Internet. Just another repetition mostly for myself.

VMware has Virtual Disk Development Kit (VDDK) which is more of an API for backup software vendors. But it includes a handy tool called vmware-mount, which gives you an ability to mount VMware virtual disks (.vmdk) from wherever you want.

Download VDDK from VMware site. It’s free. And then run vmware-mount with the following keys:

> vmware-mount driveletter: “[vmfs_datastore] vmname/diskname.vmdk” /i:”datacentername/vm/vmname” /h:vcname /u:username /s:password

Choose drive letter, specify vmdk path, inventory path to VM (put ‘vm’ in lowercase between datacenter and vm name, upper case will give you an error) and vCenter or ESXi host name.

Note however, that you can mount only vmdks from powered off VMs. But there is a workaround. You can mount vmdk from online VMs in read-only mode if you make a VM snapshot. Then the original vmdk won’t be locked by ESXi server and you will be able to mount it.

To unmount a vmdk run:

> vmware-mount diskletter: /d

There are also several GUI tools to mount vmdks. But vmware-mount is enough for me.

Consistent VMware snapshots on NetApp

If you use NetApp as a storage for you VMware hard drives, it’s wise to utilize NetApp’s powerful snapshot capabilities as an instant backup tool. I shortly mentioned in my previous post that you should disable default snapshot schedule. Snapshot is done very quickly on NetApp, but still it’s not instantaneous. If VM is running you can get .vmdks which have inconsistent data. Here I’d like to describe how you can perform consistent snapshots of VM hard drives which sit on NetApp volumes exported via NFS. Obviously it won’t work for iSCSI LUNs since you will have LUNs snapshots which are almost useless for backups.

What makes VMware virtualization platform far superior to other well-known solutions in the market is VI APIs. VI API is a set of Web services hosted on Virtual Center and ESX hosts that provides interfaces for all components and operations. Particularly, there is a Perl interface for VI API which is called VMware Infrastructure Perl Toolkit. You can download and install it for free. Using VI Perl Toolkit you can write a script which will every day put your VMs in a so called hot backup mode and make NetApp snapshots as well. Practically, hot backup mode is also a snapshot. When you create a VM snapshot, original VM hard drive is left intact and VMware starts to write delta in another file. It means that VM hard drive won’t change when making NetApp snapshot and you will get consistent .vmdk files. Now lets move to implementation.

I will write excerpts from the actual script here, because lines in the script are quite long and everything will be messed up on the blog page. I uploaded full script on FileDen. Here is the link. I apologize if you read this blog entry far later than it was published and my account or the FileDen service itself no longer exist.

VI Perl Toolkit is effectively a set of Perl scripts which you run as ready to use utilities. We will use snapshotmanager.pl which lets you create VMware VM snapshots. In the first step you make snapshots of all VMs:

\”$perl_path\perl\” -w \”$perl_toolkit_path\snapshotmanager.pl\” –server vc_ip –url https://vc_ip/sdk/vimService –username snapuser –password 123456  –operation create –snapshotname \”Daily Backup Shapshot\”

For the sake of security I created Snapshot Manager role and respective user account in Virtual Center with only two allowed operations: Create Snapshot and Remove Snapshot. Run line is self explanatory. I execute it using system($run_line) command.

After VM snapshots are created you make a NetApp snapshot:

“\$plink_path” -ssh -2 -batch -i \”private_key_path\” -l root netapp_ip snap create vm_sata snap_name

To connect to NetApp terminal I use PuTTY ssh client. putty.exe itself has a GUI and plink.exe is for batch scripting. Using this command you create snapshot of particular NetApp volume. Those which hold .vmdks in our case.

To get all VMs from hot backup mode run:

\”$perl_path\perl\” -w \”$perl_toolkit_path\snapshotmanager.pl\” –server vc_ip –url https://vc_ip/sdk/vimService –username snapuser –password 123456  –operation remove –snapshotname \”Daily Backup Shapshot\”  –children 0

By –children 0 here we tell not to remove all children snapshots.

After we familiarized ourselves with main commands, lets move on to the script logic. Apparently you will want to have several snapshots. For example 7 of them for each day of the week. It means each day, before making new snapshot you will need to remove oldest and rename others. Renaming is just for clarity. You can name your snapshots vmsnap.1, vmsnap.2, … , vmsnap.7. Where vmsnap.7 is the oldest. Each night you put your VMs in hot backup mode and delete the oldest snapshot:

“\$plink_path” -ssh -2 -batch -i \”private_key_path\” -l root netapp_ip snap delete vm_sata vmsnap.7

Then you rename other snapshots:

“\$plink_path” -ssh -2 -batch -i \”private_key_path\” -l root netapp_ip snap rename vm_sata vmsnap.6 vmsnap.7
“\$plink_path” -ssh -2 -batch -i \”private_key_path\” -l root netapp_ip snap rename vm_sata vmsnap.5 vmsnap.6
“\$plink_path” -ssh -2 -batch -i \”private_key_path\” -l root netapp_ip snap rename vm_sata vmsnap.4 vmsnap.5
“\$plink_path” -ssh -2 -batch -i \”private_key_path\” -l root netapp_ip snap rename vm_sata vmsnap.3 vmsnap.4
“\$plink_path” -ssh -2 -batch -i \”private_key_path\” -l root netapp_ip snap rename vm_sata vmsnap.2 vmsnap.3

And create the new one:

“\$plink_path” -ssh -2 -batch -i \”private_key_path\” -l root netapp_ip snap create vm_sata vmsnap.1

As a last step you bring your VMs out of hot backup mode.

Using this technique you can create short term backups of your virtual infrastructure and use them for long term retention with help of standalone backup solutions. Like backing up data from snapshots to tape library using Symantec BackupExec. I’m gonna talk about this in my later posts.

Reclaim NetApp snapshot space

Recently I needed to configure big NFS share for HPC cluster users. Proposed use of this share is research data which rapidly changes and becomes obsolete very quickly. Primary storage on site is NetApp which by default reserves large portion (20%) of volume for snapshot. They are not practical for such usage scenario, so here is the quick tip on how to reclaim snapshot reserve for users usage.

snap sched volname 0 0 0
snapdelete -a volname
snapreserve volname 0

In first command we disable snapshot creation, then delete all already created snapshots and finally disable volume reservation for snapshots.

shIT happens

After we moved our DB2 server from v9.7.3 x86_32 to v9.7.5 x86_64  (with server replacement) DB2 Storage Management has broken down. Tasks that make snapshots of tablespaces have stopped working with error:

CALL CAPTURE_STORAGEMGMT_INFO(2, ‘ ‘, ‘DATA_SPACE’)
SQL0443N Routine “SYSPROC.CAPTURE_STORAGEMGMT_INFO” (specific name “CAPT_STGMGMT_INF”) has returned an error SQLSTATE with diagnostic text “SQL0303”. SQLSTATE=38553

SQL0303N A value cannot be assigned to a host variable in the SELECT, VALUES, or FETCH statement because the data types are not compatible.

Looks like data type incompatibility. I tried dropping all SGMT_* tables, recreate them using CREATE_STORAGEMGMT_TABLES and make a snapshot. Same error.

I’ve googled everything related to this in the Internet. There are only two topics without any suggestions here and here.

It seems like this problem has no solution (IBM DB2 bug?) and we should just go with it.