Data Centers

Two data center strategies: Centralization and 'sharding'

Mary Shacklett looks at two approaches to data center strategy, which can be equally successful as long as they are tied to specific business needs.

Data center strategies are instrumental in cementing the online presence of an enterprise, but there is no no "right answer" in a data center strategy in any industry - no matter how similar the businesses are. The hospitality industry is no exception-as demonstrated by two major chains that decided on very different data center approaches with positive results. If you are in the hospitality business and you depend on the strength of online reservation bookings, how you use your data centers is paramount.

The centralization approach

Company One decided to enrich its corporate data center by consolidating staff and technology in a relatively compact geographical area. From a data center recovery and business continuation standpoint, it maintained multiple data centers — but its database of worldwide customers and reservations was monolithic. A single database for transactions was easy to maintain. It also facilitated a more straight-forward architecture when it came to feeding off the database for data warehousing and analytics. The focus on highly-tuned transaction processing and analytics that could quickly inform management where hotel room inventory was available, and how to price it for immediate occupancy, favorably contributed to hotel revenues.

From a data center perspective, centralization was a way to consolidate assets, staff and also energy utilization. This paved the way for IT savings — and also reduced risk since the data center was a company-owned asset under corporate control.

The sharding approach

Company Two became frustrated because it felt that it was losing out competitively in offshore markets. It believed that one major reason it was at a competitive disadvantage was because would-be customers were going to its website to book accommodations — but they got frustrated and abandoned the website because it took too long for them to complete an online reservation.

There was nothing wrong with the company's transaction processing engine, its database, or its U.S.-based data center. Instead, what the company was experiencing was Internet delays generated in part by poorer online service levels in some of the countries that it served, and in other cases (e.g., "the great firewall of China"), government-imposed policies for Internet traffic that slowed down transactions when they were coming in from international locations. "In these conditions, we felt that we couldn't compete on a level playing ground with hospitality providers that had their data centers in-country," said the CIO.

The company decided on a new data center approach. It abandoned its centralized data center concept, and instead began to split the main data center into a series of "mini data centers" spread throughout the different country markets that it served. To do this, it underwent a painstaking project of "sharding" its central database — a process where pieces of the database are splintered off to run independently of each other. In this case, the portion of the database that covered Chinese customers was splintered off and reinstated in a Chinese data center. A similar strategy was also undertaken for other world markets. Data center locations were leased in these respective countries.

The new data center strategy enabled the company to go head-to-head with its in-country competitors. In so doing, the company also had to deal with a new set of risks, given that the data centers were leased and in different countries, and that there was less control of staff at these remote locations. It was also more complicated to manage many different databases. Nevertheless, the strategy made sense for the business.

Finding the best fit

Today, both of these hospitality companies are thriving with their respective data center models — and in a dynamic and highly competitive industry. The fact that both are doing well with very different strategies should send a clear message to IT decision-makers: that data center strategies can be highly successful in both traditional and unorthodox deployments. The key is to keep your options open, and to seek out the "best fit" for the business.

About Mary Shacklett

Mary E. Shacklett is president of Transworld Data, a technology research and market development firm. Prior to founding the company, Mary was Senior Vice President of Marketing and Technology at TCCU, Inc., a financial services firm; Vice President o...

Editor's Picks