Massively Multiplayer Online Games (MMOGs) have recently emerged as a novel Internet-based entertainment application. Hundreds of MMOGs already serve tens of millions of paying customers world-wide, with virtual worlds such as World of Warcraft and Runescape hosting daily several millions of players. These players want fast-paced entertainment delivered through the Internet, which raises important content and resource requirements; when these are not met in full and on time, players will quit. However, the current industry approach in addressing these requirements has high cost and limited scalability. The high cost makes the market inaccessible for amateur and small game developers. The limited scalability means that even the largest game developers cannot support this rapidly growing community. We aim to investigate scalable, dependable, yet low-cost computational technologies for MMOGs.

Recently, a social component was added to MMOGs---Massively Social Gaming defines MMOGs in which the social interaction between players enhances and extends the gameplay experience, not the least through emergent behavior. We aim to investigate the social component of MMOGs, and to understand how to leverage it to make the lives of MMOG players better.

As a third main goal, we would like to better integrate gaming into education (for a complex set of reasons that include but go beyond this rationale).

An overview of the main topics and their challenges :

1. Game world simulation: scalable, epic, fast-paced, etc.
2. Game content generation: automated, crowdsourced, etc.
3. Game analytics: automated, massive-scale, real-time, persistent, durable, etc.

• Undergraduate Students: Anand Sawant, Jerry de Swart, Fedor Jutte (TU Delft), Maria Carpen, Alexandru Dimitriu, Stefan Istrate, Adrian Lascateu, Georgiana Petria (Politehnica University of Bucharest, Romania).
• Graduate Students: Siqi Shen, Yong Guo (TU Delft), Vlad Nae (U. Innsbruck, Austria).
• Development and Valorization: Otto Visser.
• Collaborators: Dick Epema, Henk Sips (both TU Delft), Thomas Fahringer, Radu Prodan (both U. Innsbruck, Austria), Nicolae Tapus, Vlad Posea, Mihaela Balint (Politehnica University of Bucharest, Romania).
• Yours truly: Alexandru Iosup.

1. What are the characteristics of real MMOG workloads? Much effort has been put into characterizing the workloads of MMO Role Playing Games, a popular branch of MMOGs. However, many questions remain unanswered before the characterization is complete. What are the differences between MMOGs and other Internet-based applications? What are the characteristics of different types of MMOGs?
    
2. How to evaluate the performance of a system, algorithm,  or mechanism designed for MMOGs? A major impediment to developing the area of MMOGs is the lack of an accepted method to evaluate the impact of new findings in system, algorithm, and mechanism design. How to evaluate the performance of MMOGs using simulations? Or real system experiments? Or analytical models?
    
3. What are the characteristics of the social networks present in MMOGs?  While researchers have investigated the structure of (social) networks for decades, the activity characteristics and the community structure of online social gaming remain relatively unknown. What are the unique characteristics of the MMOG community? What characteristics do MMOG communities share with traditional (online) communities such as FaceBook, Youtube, Microsoft Messenger, and Flickr?
    
4. Can MMOGs make use of clouds? As a consequence of their internal organization or of the policies of existing data centers towards hosting MMOGs, many MMOG operators have to maintain large computing infrastructures, either self-owned or leased over long periods of time. A possible alternative is to use resources and services provisioned from commercial clouds, perhaps as a complement to existing infrastructure and services. But which MMOGs can now leverage cloud infrastructure and platform services? Can MMOGs function well under the availability and performance profiles of the current production cloud services? Can we improve the MMOG analytics to MMOG social analysis?
    

1. Collected input to characterize the workloads of several real MMOGs. We have collected long-term traces that can be used to characterize the workloads of several real MMOGs. We have already used [1,7] one such trace to uncover the interaction between players as a major strain to the system; this was ignored in previous work on MMOG workloads, and also helps distinguishing MMOGs from well-studied Internet-based applications such as Web servers.
    
2. Designed CAMEO, a framework for continuous analytics for massively multiplayer online games (MMOGs) using cloud resources. We have provided [4] a first estimate of the feasibility and costs of performing continuous analytics for MMOGs on cloud resources.
    
3. Within edutain@grid, analyzed the feasibility of running massively multiplayer online games (MMOGs) on cloud resources. We have analyzed [1,7] the potential gains of running MMOGs on cloud resources, based on an ideal cloud/data center model. We have started to investigate [5] the impact of virtualization on running MMOGs on cloud resources.
    
4. Analyzed several aspects of the social component for the BBO / BBO Fans online bridge communities. We have analyzed [11] the community structure and the ability of the community to coordinate its activity (self-organization, synchronization) among tens of thousands of players.
    

1. "It is possible to automatically generate at massive scale puzzle instances of commercial quality on grid infrastructure." [6][7]
    
2. "Otherwise unused (spare) resources may be used to solve complex puzzle instances at almost zero cost and with reduced overhead, if appropriate resource allocation policies are used." [3]
    
3. "The performance of the production cloud services offered by Amazon Web Services and Google App Engine varies over time, and can alter significantly the performance and cost profiles of large-scale applications such as social gaming (state management, trading virtual goods, etc.)." [9]
    
4. "Continuous analytics for MMOGs on cloud resources is feasible and shows good promise of being cost-effective." [4][13]
    
5. "Dynamic resource allocation from clouds can lead to a tenfold reduction of the platform operation costs for massively multiplayer online games (MMOGs)." [1][7]
    
6. "Games can be used as and are a good vehicle for teaching the principles of Software Engineering, provided the curriculum uses the appropriate games for teaching the right topics." [1]
    
7. "Without adequate resource virtualization policies, the performance limitations threaten to cancel out the benefits of resource virtualization." [5][10]
    
8. "For online bridge, the following player behavior type are common: community builder, community member, random player, faithful player. Moreover, online bridge communties are coordinated (self-organized, synchronized) even for large numbers of players (tens of thousands). " [11][12]