What are some of the key things your clients want from their online gaming experience? According to the seminal global study released by eCOGRA in January 2007, fun and entertainment seem to rank above the profit motive as reasons for play. The most common problems experienced by players are being disconnected from the game and software malfunctions.
If the players’ goal is to have fun, and they are disconnected from the game when using your site, you can be sure their goal was not met. Unmet goals lead to dissatisfaction; and dissatisfaction leads to lost customers. Since a rule of thumb is that it costs six times as much to win a new customer as it does to service an existing customer, losing customers has an immediate, large and long-lasting impact on gaming operators’ bottom lines. With the numerous chat rooms and sites, the impact is worsened through negative word-of-mouth discussions if disconnected sessions occur with any frequency.
Now that the main complaint has been voiced, it's up to the gaming operators to figure out how to address the problem. That’s where your computer infrastructure solution fits in. There are only three ways that a customer can be disconnected from a game.
1. The problem is at the customer’s end. Their computer hangs; their Internet Service Provider (ISP) drops the connection; they close the session by mistake--the list is endless.
2. There is a major, catastrophic interruption on the Internet. This is rare, but can happen. In January 2007, an earthquake devastated some fiber optic cables beneath the sea off the Taiwanese coast, and a week later, Internet services were still not fully recovered. This isn’t a new problem--do you remember back in 2001 when a fisherman accidentally severed an Internet submarine cable, interrupting access to U.S. Web sites by users in mainland China and Taiwan? Failures of this sort are considered beyond the scope of any online gaming operator to solve, but the next area is definitely within their control.
3. There is a problem at the operator’s end. A server hangs without adequate switchover; the bandwidth source is compromised, without having redundancy built in to maintain Internet access; there is a power outage and generators are undersized to keep everything operating properly; there is a power surge and electronic components fail … This list, too, is endless, and can be compounded by a problem of a different nature: Your business has been growing so quickly that your infrastructure just can’t keep up with the client demand. For anyone who has ever landed at a busy airport at 5:00 p.m. and tried to get a cell phone signal to call home, you know what I mean. Blackberry users might remember the network outage that caused widespread service disruptions across North America overnight between April 18th and 19th.
Problems at the operator’s end are addressable by operators. For problems involving explosive growth, adding transaction processing capability by adding more servers works, if your software solution has been designed for growth, as enterprise systems are. For solutions with limits on database files, such as a system allowing a maximum of 50,000 customers, extensive work may be required to address the challenge, which could entail moving onto enterprise solutions such as those using Oracle databases.
For operational problems involving power and bandwidth, the problem and solution often lie with your computer room setup. The cost to provide adequate, fully redundant bandwidth becomes prohibitive for computer operations located outside of a data center. By contrast, the shared infrastructure of a hosted data center is a much more economical way to maintain access for customers, with minimal games dropped. Not all data centers are created equal, however, so you need to check yours out fully to ensure it has enough power, cooling, bandwidth and conditioned space to serve you now and in the future. (More on the need for scalable growth in a future article!) In any event, as you look at various data centers for hosting your computer infrastructure, things to check for, in terms of reliability, availability and scalability include:
1. Plenty of redundant, dedicated bandwidth. Redundant means that in the event of a problem with one of the bandwidth providers, (e.g. cable got hit by a backhoe), your Internet traffic will automatically and seamlessly be routed through your alternate bandwidth provider. Dedicated bandwidth means that the capacity is reserved for your use only, so your customers aren’t left hanging while other data center customers are chewing up the available capacity. A data center offering a few hundred megabits of bandwidth will likely run into problems satisfying all of its customers. Inquire after at least a gigabit of redundant bandwidth from your provider.
2. Lots of available power. If your potential data center is well known to the local power station--that is, if your center is a significant user of available power in the neighborhood--there's a problem. You want to know that doubling, tripling or increasing your power several hundred percentage points over time isn’t going to require the local power station to build a new generating facility. With the way that computer power consumption is growing, your ideal data center should constitute little more than a blip on the local power station’s radar. It is expected that power consumption per 1U of server will quadruple over the next three years; and if today’s trends continue, power consumption could grow to over sixteen times its current level over the next six years.
3. Space to grow in. Your data center needs to have available, expandable space, both now and in the future, so that you won’t need to switch facilities. Moving is disruptive--far better to grow within your existing facility than endure a forced move because of space, power or bandwidth constraints
4. Adjustable cooling, and lots of it. Computer equipment runs efficiently when it is adequately cooled, and servers reduce their clock speed by up to 50 percent when they run hot. That could mean you would be purchasing twice as many servers as you need! Since computers have a habit of running hotter the faster the clock speed, it is expected that cooling will need to adjust over time. At least 160 CFM per kW of power consumption is needed to protect modern computer equipment. In-row cooling is best for handling blade servers and new technology; it can easily be augmented to increase cooling as needs change.
5. Ability to monitor the online gamers’ experience, along with automated trouble shooting to help identify bottlenecks. This ability is the crucial to proactively managing the end-customer experience, to fixing your infrastructure dynamically before online gamers even notice a slow-down. And doing this without the gamer needing to download any software to their machine makes the process seamless and unobtrusive. This functionality does exist, and is a leading-edge service that can mean the difference between intrusion and uninterrupted playing for increased revenues.
As you can see, there is plenty that you can do to address well documented problems among gamers, and with a little due diligence, you can increase your service levels while effectively managing your costs.
eNation Corporation's latest project was to design, build and manage a 25,000-square-foot, state-of-the-art hosted data center on behalf of Alexander Internet Technologies (AIT), located in Edmonton, Alberta, Canada. With fully redundant power and up to 10GB+ redundant bandwidth, in-row cooling able to handle standard through-grid and high-density computing and adept end-user experience monitoring and troubleshooting systems, AIT is well positioned to assist you.
Future articles will discuss additional elements to consider when selecting your hosted data center, including: service and support; security; adequate backup generators; early fire detection and suppression; redundancy to ensure no single point of failure; service level agreements; easy access to a highly skilled workforce; stable weather (monsoons and data centers don’t mix well); geopolitical environment; and associated issues regarding licensing jurisdictions and regulatory environments.