Paul Morrissey

Automating VM Snapshot Backups and Recovering 1TB VMs in seconds with Hitachi Virtual Infrastructure Integrator

Blog Post created by Paul Morrissey Employee on Oct 12, 2014

One of the key areas we assist our customer's leveraging HDS Storage for their VMware environments is automated VM-level backups. This is one of the key areas addressed by a solution called Hitachi Virtual Infrastructure Integrator (known internally as V2I).. As the product slide boldly claims, It provides managed data protection services for server VMs and desktop VMs integrated with VM level hardware storage snapshots, all accessible  by VMware administrator. One of our clients was discussing their use case using the product the other day and I thought I'd share more broadly so others can benefit.


Environment and use case: They have a large VMware farm (1000+) and needed to move beyond their traditional host OS based backup to provide better levels of recovery services while managing their backup storage growth. Specifically, they wanted to ensure that every VM deployed had an automatic data protection policy allowing a minimum recovery option. Too often VM's would get deployed and backup app enablement would be missed with no recovery option available which led to some red faces. They also wanted to give the app groups a certain level of self restore capabilities to provide secure flexible access to their backups so app owners could perform necessary self-restore directly without dragging backup data across the network. They also had some large VMs that they were also concerned about effective recovery times if something went awry.

Solution Summary: Deployed Hitachi Virtual Infrastructure Integrator configured to manage a 2 node HNAS 4100-HUS VM storage platform. They created 2 backup group schedules for each ESXi cluster, with minimum of 4 snapshots per day; Every VM deployed onto the cluster inherited that backup schedule and they now have 4 recovery points each day for every VM in the environment. When an application group needs to access a previous snapshot of their environment (e.g. access a copy of their MS SQL or mySQL database from last Monday before 3pm), the VM admin/operations can one click an operation to have that snapshot mounted directly to the VM (e.g X: drive in windows for example). With Virtual Infrastructure Integrator enabled file clone snapshots offloaded to the hardware accelerated HNAS 4100-HUS VM platform, 1 TB VM are restored in 5 seconds. That is Telsa Model D speed..


Ok, so how does Virtual infrastructure Integrator (V2I) enable this. Below is the architecture. It runs as a VM appliance, managed through a vSphere plugin. The V2I VM appliance has API access to HDS Storage (in this case NFS datastores served from Hitachi NAS (HNAS)) in order to execute VM and VMDK level snapshot operations. ]

V2I diagram.png


Within the HNAS 4100-HUS VM system, The VM snapshots are space efficient clones at the file level and are independent of each other from a recovery perspective. Looking at screenshot below, when we do a VM level snapshot with Virtual Infrastructure Integrator, it will selectively snapshots just the individual VMDKs (and .vmx/.vmf).  These are 0 bytes snapshots and only consume delta space when changes to that individual VMDK are changed. This is one of the reasons the HNAS platform can support up to 100 Million snapshot clones and support more recovery points unlike classic NAS devices which track changes at a volume level

V2I-VMDK level snapshot.png

With the Hitachi vSphere plugin that is provided with the software, the customer created a scheduled based snapshot backup. The screenshot below is for datastore but you have the flexibility to do this at a VM, datastore, vApp and custom group level. Any time that a new VM gets deployed into that container (e.g datastore, vApp or custom group), it will be automatically be backed up per that container backup schedule

V2I-sched backup.png

When backups are completed (even heavy production hours), each VM will surface in the plugin the list of VM storage based snapshots that have been created for it. You'll see from the screenshot below that we list the VM snapshots by age and it provides various options you can act on those snapshots.. (e.g. Restore directly, Fast clone copy of the snapshot, delete and most interestingly for this client, Mount the snapshots).




When the VM operations group gets a request for access to previous backup, they simply go to the interface, identify the snapshot in question and click "Mount Snapshot Backup VMDKs". "Like magic with the snapshot disk appearing in the production VM.. Neat and efficient.."


This video snippet below shows the above use case in operation with the product {I skipped adding a voiceover, suggest click to watch on youtube, 720p}.  As you will see, the SQL VM has already scheduled based backup managed by Virtual Infrastructure Integrator in place and the VM admin has also chosen to do an additional manual backup (called emergency patch") before applying some new patches to the environment.  Instead of restoring the full VM, they need more granular recovery of the data that was on E drive (SQL DB). The video shows how a snapshot backup copy of the MS SQL database is directly mounted to the production SQL VM as another disk (in this case, an F: Drive) with single click; This enables the App Owner to perform their own level of recovery (drag/drop from F disk) or compare instances directly. Importantly, we are directly accessing Snapshot VMDK over storage network as opposed to slower access to network share backup....  The flexibility and speed was the compelling value that this particular customer attributed to this solution in their environment..




> For closer look, you can check it out and get hands on with it at VMworld Barcelona 2104 at Hitachi P205. If not in town and want 1-on-1 with colleague, drop your name here


Stay tuned for other updates on the capabilities of Hitachi Virtual Infrastructure Integrator.