Peer-to-Peer

Server consolidation, virtually

Macawber Beekay found that deploying the VMware Infrastructure has brought down the maintenance and support costs while delivering improved uptime, easier manageability, greater flexibility and enterprise-class backup, writes Nikita Upadhyay

Macawber Beekay deployed the VMware Infrastructure 3.5 to vein their data centers with a view to facilitate data center consolidation. The first PoC (Proof of Concept) was done in February 2007 using the parallel setup in their primary data center. The company found that the backupand upgrading tasks required to ensure continued operation of its physical infrastructure, were hampering its ability to deliver always on IT.

The complications

Before the deployment, the company was not using any virtualization technology. It had a hard time when it came to hardware upgrade, patch management, security update, software upgrade, etc. With no high amenity and feature, these processes were time consuming and complex. Realizing that and with a view to reduce downtime and for better ROI, it moved towards implementing the VMware Infrastructure.

The ever increasing cost of power and cooling requirements acted as the catalyst towards the decision. The ever scaling cable complexity and space requirements as well as cost additions of backup client licenses made them move towards the virtualization technology. The increasing time required for deployment of hardware, firmware upgrade, efforts for deploying new machines and servers and the right time to give a boost to IT frame added to the urgency for the deployment.

Phases of implementation

When Nishant Sinha, Senior Manager, IT, Macawber Beekay, found himself dealing with a growing hardware maintenance task that compromised his capacity to deliver ‘always available’ information technology services, he turned to VMware virtualization. “For us to perform the required server maintenance, we had to shut down each server for two to four hours a week while our weekly full LAN backups had to run for more than 48 hours at a time,” said Sinha.

The implementation was carried out in three phases. The first phase consisted of implementation of IBM x3650 servers. This was done to facilitate NFS configuration. Two of these servers were added to promote Network File Sharing system which is a common storage system.

The second phase consisted of adding High Availability (HA) and VMotion capability (which means automatic shift of virtual servers from failed physical server to another working physical server). The final phase consisted of adding two IBM x3800 series quad processor servers to VMware Infrastructure. The implementation ended with the final transition of all the physical servers to the VMware Infrastructure. The entire deployment took approximately one month for its accomplishment.

Working mechanism

ESX servers provide a common HAL (Hardware Acceleration Layer) across all guest OS, supported by VMWare tools for specific guest OSs. “Our production environment now contains four ESX servers. Capacity planning is done against available resources and then an analysis of the physical machine is performed using consolidation feature of VMware Infrastructure server.

The deployment supports 54-bit computing and is capable enough to support 64-bit computing. “Our centralized virtual backup connected to a SAN across a fiber channel allows us to complete LAN backups in less than four hours without switching anything off. Our application and services availability is now up to 99.99%,” informed Sinha.

The initial investment was approximately Rs 8 lakh, which showed good ROI and Total Cost of Ownership (TCO). Savings in terms of manpower, electricity cost and cooling supersedes the investment done. A Few OS dump and Blue Screen Of Death (BSOD) was encountered initially by Physical to Virtual (P2V) instances because of hardware changes, but it took minimal time to stabilize all systems on VM infrastructure after installing latest VMTools.

Benefits of the deployment

The company now experiences much flexibility in any new deployments within the data center. It has also observed reduced outage due to system restarts during the hardware and firmware upgrades. Template based deployments have brought down the implementation time by 70%. The cost has been cut down by 30%. The company has experienced high availability with low downtime.

VMware Consolidated Backup (VCB) integration with Legato eased the backup procedure and reduced backup window. Physical servers were reduced and four virtual servers were added, which impro-ved the manageability. Data center density is contained, thus reducing power and cooling requirements.

“Our centralized virtual backup connected to a SAN across a fiber channel allows us to complete LAN backups in less than four hours without switching anything off. Our application and services availability is now up to 99.99%,” stated Sinha.

By consolidating 15 physical servers to an ESX server environment running across four production hosts and one backup host, Macawber Beekay has implemented a flexible, dynamic infrastructure that can scale to support growth. The organization plans to enhance its backup and disaster recovery capabilities by extending the virtualized environment across a second yet-to-be commissioned data center. The disaster recovery solution that the company is about to implement is still in the planning phase.

nikita.upadhyay@expressindia.com