A stitch in time

LG Electronics’ disaster recovery site will help prevent prolonged downtime and loss of data, which could have a dual impact—monetary loss running into several crore rupees as well as lost business opportunities, says Abhinav Singh

The recovery time for critical information is as low as four hours now and that of sensitive information, eight hours, says Daya Prakash

LG Electronics India (LGEIL) has become a household name in the Indian electronics market. It has a huge presence in the country with two manufacturing plants—one located in a 47-acre property at Noida near Delhi and the other in Pune. Powered by its dizzying growth in the last two years, LGEIL has witnessed a doubling in data capacity every year. The company’s infrastructure has kept pace with its data growth from 30 branches in early 2004 to 43 by year end. The company expects its stock points to increase from 50 to 65, and the number of area offices from 74 to about 100 by end 2004. Additionally LGEIL has approximately 3,000 internal users generating huge amounts of data. The company expects to touch 280 GB of data by end 2004. Ever since LGEIL’s Pune plant came up in June 2004, it has been running a disaster recovery operation between its Pune and Noida plants.

Disaster recovery (DR): a must for business continuity

LGEIL has seen tremendous growth, totting up revenues of Rs 3,300 crores in 2003-04. That’s expected to double in 2004-05 to Rs 7,000 crores. Daya Prakash, project manager (LG Electronics India - ERP & eBiz) LG Soft India, explains, “The huge growth in revenues which we are expecting this year translates into Rs 20 crore worth of sales every day. It can be further broken down to Rs 2 crore of sales every hour. Downtime of even 2 to 4 hours would have a severe impact on our bottom line reflecting a loss of Rs 4-8 crore in revenue. This makes it important to have a DR site so that business critical data can be protected and backed up in safe vaults.”

Prakash explains that since LGEIL’s sales targets grew by almost 100 percent there was a need to increase production capacity. To achieve its required production, LGEIL decided to set up a second DR site in Pune as the existing DR site in Noida was unable to meet its requirements. While planning for the IT infrastructure at Pune, LGEIL decided to replicate the Noida model for DR as well as business continuity requirements. The DR site in Pune uses Sun’s V880 and Sun Fire 6800 servers while at Noida Sun Fire 6800 and Sun Fire 4800 are used.

LGEIL had kick-started its DR initiative in December 2002. It already had a cold site running at Noida wherein the backup window for 65 GB of business critical data (such as ERP related data) was three hours while the recovery time was as long as three weeks. The cold site had very basic infrastructure in which there were two production servers (Sun Fire 4800 and V 880) and backup was done manually onto a HP LTO tape drive.

In February 2003, LGEIL reached the status of a warm site with recovery time reduced to just one week for 142 GB of critical data. The warm site was facilitated when LGEIL went in for additional hardware in the form of a backup server (a Sun V880). Currently LGEIL has reached the status of a semi-hot site wherein it has the requisite DR infrastructure at its site in Pune. The recovery time has further come down to as little as a day. The semi-hot status means that the data is physically stored on tapes and is also transmitted to the other location (from Noida to Pune and vice versa) by File Transfer Protocol (FTP). Presently, two sets of tapes are backed up at the two DR sites on a daily basis. One set of tapes is kept in fireproof vaults whereas the other set is kept at LGEIL’s branch office, 35 kilometres away from the main DR site in Noida, in a HP DDS-4 tape library.

When servers fail

On a normal day, Noida’s (plant and corporate) users are connected to the Noida server and the Pune (plant) users are connected to the Pune server. If the Noida server fails all the critical users–both plant and corporate–would be connected to the Pune server to execute critical activities such as sales and production. A similar connection to Noida is made if the Pune server fails. Prakash says, “We have classified all the information into two categories–critical and sensitive. (Critical data refers to the ERP and business-related data while sensitive data includes all e-mail, Excel sheets and PowerPoint presentations) The recovery time for critical information is now as low as four hours and that of sensitive information is eight hours.”

Save our sales

LGEIL is prepared with the initiation of its DR implementation to meet any unforeseen event leading to disaster. A proper DR set-up in place can save the company several crore rupees. LGEIL would also not miss business opportunities because of critical data being unavailable for decision-making.

Prakash says, “Our DR process has helped us in achieving our turnover targets and has streamlined the time to market. Without a DR process, downtime can reflect negatively on the manufacturing process, which can affect our sales.”

Mirror, mirror on the wall

LGEIL will aim at migrating to a hot site (always-on DR service) status from the current semi-hot site status during the first quarter of 2005. The migration to a hot site would require LGEIL to start using data mirroring so that information can be replicated at both sites. Data mirroring is already under evaluation and would facilitate a parallel processing operation. LGEIL plans to implement Data Guard from Oracle and Sun’s StorEdge Availability Suite’s point-in-time copy software for the mirroring. Once data mirroring is implemented, recovery time is expected to be as low as four hours.

abhinav@expresscomputeronline.com