While virtualized environments are the rage today, virtualized environments also suffer from I/O bottlenecks as servers can get inundated with I/O requests from multiple virtual machines. I/O congestion is caused due to several physical servers rolled into virtual server instances on a single physical machine with a single CPU, memory, storage and network card.

“Assume that you consolidate 15 physical servers into one physical server with virtualization. Now you have a single CPU, memory, NIC card, storage device, handling 15 virtual servers. Due to this, the server gets choked with I/O requests. To ease the I/O operations, most organizations go in for a consolidation ratio of 8:1, though they can achieve a ratio of 15:1,” says PK Gupta, Director and Chief Architect, BRS Practice-APJ, EMC.

Hence, while the actual physical servers are consolidated at the virtual level, data redundancies exist at the physical storage level. This means increase in storage, processing and bandwidth requirements when backing up or restoring these virtual images. As de-duplication reduces data before it gets on to the network and reduces bandwidth requirements, EMC is finding huge demand for its de-duplication technologies in virtualized environments. 

“Source based de-duplication can handle issues of I/O congestion in virtualized environments. The solution scans every virtual image to check for redundant data of operating systems or applications and de-duplicates this data. This de-duplicated virtual image data is then sent for back up. For the next back up, only that block of data from each virtual machine file, which has undergone a change is sent for back up,” says Gupta.

This ensures that there are no I/O bottlenecks from the backup and recovery point of view. It also reduces the amount of storage required to back up the virtual server data and the bandwidth to send these virtual images over the wire.