VMware’s Virtual Machine File System (VMFS) uses clustering to provide improved performance and resilience in an enterprise environment. So what are the pros and cons of the file system, and what are the implications for VMFS recovery?
What are the benefits of using VMFS?
Using the VMFS format, multiple virtual servers can read and write to the same file system simultaneously which would be prevented by file locks in traditional disk systems. This enhanced read/write capabilities then means that data stores can be consolidated, reducing the physical hardware required for server farms.
The clustering approach also helps to improve the overall resilience of virtual servers, allowing data to replicate between host servers and disk arrays to prevent problems should one node fail.
VMFS clusters (like other virtualisation technologies) allow you to consolidate data and maximise resource usage, ensuring that none of your existing capacity is wasted. And as you would expect, virtual machine images and data can be deleted without affecting other servers.
What are the drawbacks of using VMFS?
One of the greatest strengths of VMFS is also one of its greatest weaknesses. Data consolidation helps to reduce capital and operational expenditure by rationalising the physical hardware used in the data centre. Although this reduces potential points of failure, data consolidation using VMFS also reduces redundancy, increasing the potential impact of any failure and making data recovery more complicated. The inherent complexity of VMFS makes it perfect for synchronous read/write operations, but a nightmare in the event of hardware failure.
There are also specific issues with earlier versions of the VMFS file system. Only by using VMFS version 5 in addition to vSphere 5.x can you reach the heady heights of 64TB extent limit. Users of vSphere 5.4 or lower cannot create VMDKs larger than 2TB, regardless of the file system version in use for instance.
Another drawback comes in the lack of data de-duplication functionality built into VMFS. This means that storage space is wasted both inside the cluster and in each backup. The cost of storage is still continuing to fall, but duplicate data storage is an avoidable waste of budget.
VMFS recovery – is it even possible?
Although complex, it is possible to recover lost, deleted or corrupted files from VMFS volumes. Obviously the preferred method will be direct from a recent backup. But if that is unavailable or unusable, more in depth VMFS recovery tools and methods are required.
Ultimately, virtual infrastructures saves money and maximises resources, but also creates additional complexities when it comes to backup and recovery. To help reduce storage requirements, a backup solution that provides data de-duplication may be preferable, although this could slow the VMFS recovery process.
It is also important to prepare for the eventuality that backup provisions may also fail, particularly where they share the same physical infrastructure as the VMFS cluster. To reduce costly downtime your backup provisions should also include access to a third party expert VMFS recovery service with the knowledge and experience required to rebuild the file system and harvest the “lost” data. Being prepared for the worst will help to reduce downtime if and when something does go wrong with the data store.