What It Takes to Build World’s Third Largest Datacenter

Summary:

From four people working as software resellers to becoming one of India’s largest telecommunication services company–Tulip Telecom has come a long way. The company recently built Asia’s largest and world’s third largest datacenter.

 

Highlights:

Achieving a PUE of 1.8 – difficult but not impossible

Reader ROI:

What PUE means to datacenter costsGreen IT and efficiency go hand-in-hand

Organization: From four people working as software resellers to becoming one of India’s largest telecommunication services company–Tulip Telecom has come a long way. The company recently built Asia’s largest and world’s third largest datacenter.

The Problem: Tulip Telecom’s move to build Asia’s largest datacenter had its fair bit of challenges.

With over 500 crores being spent only on supplying 80 MW power to the facility, the company couldn’t afford any power loss. They decided to connect to the 66KV national grid and tap into highly available power. With a datacenter housing 16,000 racks backed spread across an area of 9 lac sq. ft, the power supply and its operation required effective power usage for efficient performance, usually measured in the industry with PUE (Power Usage Effectiveness).

 “The average PUE of datacenters worldwide is found to be 1.8, according to the Uptime Institute in the USA,” said. S Rajagopal, Former CEO, Tulip Telcom Data Centre Services. “And what we achieved at Tulip is 1.5.”

Establishing an expansive datacenter spreading to 9,00,000 sq ft with a capacity to house up to 16, 000 racks asked for stringent delivery practices. “It’s very important to gain global acceptance while building a large facility as we address both domestic and international customers. Understanding global standards is the key,” said Rajagopal.

It’s very important to gain global acceptance while building a large facility as we address both domestic and international customers.

Solution: Rajagopal began with learning from example. He visited more than 60 datacenters worldwide to help with his research before drafting the design with IBM. Tulip also considered valuable customer feedback that IBM gained through customer service and planned for the scope of the project. “When addressing a very large base of customers, adhering to global standards do not allow much flexibility to experiment. You must know your core deliverable, which in my case, were a high level of security and uptime – and of course power efficiency,” explains Rajagopal.

 “We then installed a step-down transformer very near UPS.  It tremendously reduced power loss during transmission and eliminated chances of power theft and brought in 5 percent reduction in transmission loss,” he said. That’s a significant saving of 25 crores.

Under the contract, IBM will be providing consultancy services for 4,00,000 sq.ft. of DC space along with the execution of first phase of datacenter built to the tune of 20,000 sq ft.  IBM used its Enterprise Modular Data Center (EMDC) design for Tulip, to deliver a comprehensive set of capabilities that matches the highest quality standards in the enterprise services domain that are available in the world today. IBM provided the complete turnkey solution for building this DC from supplying all products like chillers, DG sets, UPS, Fire systems, to build this data centre.

It’s very important to gain global acceptance while building a large facility as we address both domestic and international customers.

Tulip addressed the facility’s cooling requirements by bringing in greener practices. “The datacenter requires a total cooling of about 16000 tons and implementing only water-based solution would cost me 27 lakh liters of water everyday. Considering that Bangalore’s climate enjoys a 15deg.C for 2000 hours per year, this was a very viable option,” said Rajagopal. Benefits: Tulip achieved power efficiency through steadfast commitments to improvement, “Efficiency to the scale of 1.5 is not exceptionally good, but for a data center of this magnitude, it is pretty commendable,”says Rajagopal.  PUE is the measure of energy used by the data center– a PUE of 1.0 depicts a perfectly efficient data center.