In The Trenches: Actifio Data Capture Methods
Part 1 – Data Capture Methods
Out-Of-Band Integration Options
When leveraging out of band integration, the Actifio application captures data and takes an application consistent snapshot using 1 of 4 different techniques:
- VMWare vStorage APIs
- RMAN for Oracle
- VSS for Windows
- Actifio Connectors for UNIX/LINUX and Windows
Actifio follows several steps to create the consistent snapshot of the data:
- Actifio quiesces the application
- Takes a snapshot through the vStorage APIs from VMWare, RMAN from Oracle, VSS for Windows, or by using the Actifo Connectors for UNIX/LINUX and Windows.
- Unquieses the application.
- Creates an incremental update of most recent image from snapshot
- Delete snapshot virtual full is contained within the Actifio CDS.
In-Band Integration Steps
In-Band Data Capture is the 2nd broad category of techniques that can be leveraged.
The advantage of In-Band Data Capture is that it can be done much faster when compared to an API call. The process is much like a LUN level snapshot using EMC’s TimeFinder or HDS’ ShadowImage.
Think of the In-Band data capture as the pass-through of the LUN with no manipulation of the data.
- LUNs are remapped from the array to Actifio Appliance and back to the Server Host.
- Quiesce application via Actifio Connector
- Snapshot in-band LUN
- Unqueisce the application via Actifio Connector
Data Capture Steps:
- Quiesce Application – Actifio connector will Quiesce
- Snapshot In-band LUN. The snapshot can live in the snapshot pool. Pointer based snapshot lives in the snapshot pool, or we can take that snapshot and write it directly to dedup for longer term retention.
VMWare API Data Capture
Protecting a VMware or Hyper-V virtual machine farm using an API call is simple and straightforward. Here are the high-level deployment steps:
- Issue an API call to vCenter to take a VM snapshot. Could be 1 VM, or more, depending the definition of the Actifio SLA. VMware tools will quiesce the VM to give us a consistent snapshot. Actifio leverages VMtools to do this completely agent-less.
- Query changed block tracking (CBT) to List Changed Blocks. What blocks have changed since last snapshot? Get a list from VMWare of the changed blocks.
- Copy incremental changes
- Out-of-band via Ethernet to Actifio appliance, or
- In-band via SAN
- Apply changed blocks, Snapshot on Actifio snapshot pool.
- Delete VMWare snapshot.
Oracle Data Capture Method (RMAN Integration):
Regardless of where the database’s storage lives, Actifio can capture the data using its RMAN connector.
- First, we take a staging disk from the Actifio appliance and mount it to the host as either FC or iSCSI.
- Next, we configure that database host to run the Actifio RMAN connector.
- The connector then uses the RMAN utility to create a copy (level 0) of the database. Actifio will incrementally update staging disk with RMAN image copy, and uses block change tracking, to apply changes to the DB copy on staging disk.
- Then, Actifio will unmount staging disk on server
- Lastly, Actifio will create a snapshot staging disk on the Actifio appliance.
Now that you have captured your data, you need to do something with it. Perhaps you want to present it to a new VM, an existing VM, or a physical host. Because this data is stored in its raw application format, no translation is needed. We do not need to access the Backup software agent or restore it to a media server. Data is instantly available from the snapshot pool regardless of size. Actifio will not move any data, rather it will access and open the required blocks during the mount operation. It knows what blocks belong to what applications, what is a snapshot copy compared to a de-duplicated copy, what is local, and what is remote.
Say for example, we had a 5TB Oracle database that we needed to recover, mount, open the database, and issue some SQL statements against to prove the data is up to date and within our Recovery Point Objective. Actifio can determine that we do not need to re-hydrate the entire 5TB database. Rather, it will calculate that we have 95% of our data in the snapshot pool, maybe I need to rehydrate 2% from the local data pool, and rehydrate 3% from the dedup pool. The Actifio controller will create separate operations for that underneath the covers which results in speed, less storage used, and operational simplicity.
Actifio can also be configured to open the database, log-in, and to issue some SQL statements against the tables to verify data integrity and synchronization.
In part 2 of this series, we will take a closer look at recovery for Oracle database environments.