BFSI VISION | Virtualization


Joydeep Dutta

Chief Technology Officer, Central Depository
Services (India) Ltd. (CDSL)

Virtualization in Core Depository Application

Adoption level and goals over the medium term

CDSL has implemented server virtualization for its critical software applications including the core depository application.

Our Goals were to achieve

  • Server consolidation, and thereby better utilization of hardware capacity and processing power
  • Reduction in data centre footprint, and thereby saving in power and cooling costs
  • Cost reduction in software licensing

Benefits of virtualization
The goals as enlisted in the previous question are also the benefits

Additional benefits have been:
  • Faster server provisioning
  • Quick server failover in case of outage in primary server
  • Improved disaster recovery process and speed
  • Overall better centralized administration

Elements that go into making a virtualization strategy

Before adopting virtualization, we need to check if the applications are compatible in virtualized environment. Proper sizing of the individual applications need to be done to plan which applications will go with which server, keeping in mind that the peaks of the applications are not at the same time. Critical applications should not be co-existing with non-critical applications. If not planned properly, using several virtual machines on a single physical server can increase server resources utilization. For example, I/O bottlenecks can occur when multiple VMs on a single server share the same network card, especially in I/O-intensive applications. This situation can contribute to degraded network performance and increased response-time latency. Network interfaces also need to be properly sized.

Challenges faced

We have seen that if base operating system is not stable and secure, then underneath VMs are also not stable and secure. In virtualized environments, troubleshooting is sometimes difficult to conclude root cause of issues. Server virtualization software lacks information about application awareness and its resources such as network, storage and database usage that is often vital for making effective load-balancing decisions.