SAN

A SAN in the NIC of time

Information generated over time at the National Informatics Centre (NIC) increased at a rapid rate. With its heterogeneous mix of servers, NIC could not fully utilize its storage. So, NIC opted for a SAN solution to ease storage management and lower administration cost. by Soutiman Das Gupta

The National Informatics Centre (NIC) was faced with a difficult situation. As the IT solutions providers to the government of India, NIC was faced with a brow-crinkling problem. Its data volumes were increasing uncontrollably, like bacteria growing wildly as a result of a scientific experiment that had foundered disastrously. It just had to devise a new way for storage management.

So NIC deployed a SAN solution. It is now able to administer and manage the vast amount of storage in the organization. It found that the SAN helps reduce maintenance costs, and makes the company's information highly available.

NIC's massive storage problem

NIC provides solutions for the IT needs of the government of India at all levels. The organization offers a range of services, including digital certificates, computer-aided paperless examination systems, development of e-Governance applications, hosting of websites, and other network services.

NIC is also designated as the 'powerhouse' of e-governance

initiatives in India. In addition to its e-governance operations, the organization's services range from consultancy, software design and development, office automation and networking services to training, video conferencing, CAD, EDI, multimedia and Internet services.

Besides these offerings, jobs related to support—such as a range of services across a large number of government organizations and departments—generate a huge amount of data. And, such a large base of information requires to be stored, retrieved, archived, entered into databases, managed, monitored, kept secure, and backed up.

The company set up a satellite-based nationwide communication network called NICNET, with over 1,400 nodes inter-connecting the national capital, the state capitals and the district headquarters in India. Each of these nodes acted as points of creation or information and many were vital for business.

DAS proved insufficient

The organization used servers with DAS-based storage to store the critical information. But the daily management, storage and backup of the vast amount of data proved quite a daunting task.

“For each individual server with DAS, our IT staff needed to perform disk health-check and data backup on a daily basis. From time to time, we also needed to conduct disk and tape storage planning to anticipate for rapid data growth,” said Dr. N. Vijayaditya, Director General of NIC.

“Whenever a new server ran a new application, our workload just doubled because data storage and related administrative tasks were not centralized,” he added.

Resource utilization suffered

While the workload increased with the growing number of servers and software applications, resource utilization headed in the opposite direction. As disk storage capacity needed to be estimated whenever a new e-governance application was launched, it was not surprising that in many instances planned storage spaces were not fully utilized.

“Some applications entail ever-increasing disk space consumption, while others might just have very limited growth. Allocating the excess storage space from one server to another, however, was not practical due to the lack of bandwidth. Without a high speed network, shared disk space over a LAN is of limited use, particularly for data-intensive operations,” explained Dr Vijayaditya.

“We were having more and more servers, and inevitably the cost-per-storage unit ratio increased continually,” he added.

Heterogeneous environment

The storage management challenge was even greater since NIC's data center functions in a heterogeneous environment. Among the 46 servers that form a block troop of computers, the platforms included Microsoft Windows 2000/NT, Linux and Sun Solaris.

Since data storage administration differs from platform to platform, a variety of support skill sets were required. Deploying a staff force for each skill set made the implementation inflexible and expensive, according to NIC's Technical Director R.S. Mani.

The demand for 24X7 up-time further complicated matters. Given the number and variety of hardware and software utilized, it was difficult to meet this objective cost-effectively.

SAN to the rescue

To overcome the challenge, NIC decided to deploy a SAN solution. This would imply better control over the systems and protection of the integrity of the data. It would also allow easier backup and failover strategies.

NIC's SAN architecture comprises four Brocade SilkWorm 2800 switches, with HDS 9960 for online disk storage and IBM LTO-DL/40 tape drives for offline backup.

Interconnecting with NIC's 46 servers, the dual fabric SAN forms a high-speed storage network that immediately cures the inadequacy of the original DAS devices and provides resilience capability against storage device failure.

“We didn't look at any other switches as Brocade's were the default option provided by most storage vendors,” said Mani.

Operational efficiency benefited

The SAN solution helped to lower administrative and management costs, raised resource utilization, and provided fault-tolerant data storage and access without compromising on security and performance.

“With the SAN in place, we're able to reduce our administration work through storage consolidation to just a few disk and tape storage systems, which are centrally managed by software from HDS. Now the efficiency of data management, storage and backup is much higher, and administrative costs have gone down significantly,”said Mani.

“Resource utilization has also improved. With disks and tapes centrally accessible through the SAN, wastage of excess storage space is reduced to a minimal, resulting in considerably lower cost-per-storage unit ratio,” he added.

The company claims that different platforms often come with incompatible high-availability technologies. Prior to having a SAN, implementing high-availability features for direct storage devices for these servers was no easy task. With SAN however, all the resilience requirements demanded by the system are either in-built, or available by simple configurations.

Future course

Impressed with the immediate success of the SAN, NIC plans to expand its infrastructure from the current single, centralized SAN to multiple SAN islands.

And in order to facilitate this, the company plans to deploy more Brocade SilkWorm switches. It also plans to build a disaster recovery center to cater to physical-type disasters like floods or fire.

Solution highlights

The solution is a dual fabric SAN assembled from:

• Four Brocade SilkWorm 2800 fabric switches
• Hitachi HDS 9960 disk storage
• IBM LTO tape drives, running HDS management software, and Brightstore Backup software

Soutiman Das Gupta can be reached at soutimand@networkmagazineindia.com