**7.3 Resources**

During the 2010 data taking the available resources were sufficient to cover the needs of experiments, but during 2011 the computing slots as well as the storage capacities at sites started to be full. Since the experience clearly shows that delivery of the Physics results is limited by resources, the experiments are facing a necessity of more efficient usage of existing resources. There are task forces studying the possibility of using the next generations computing and storage technologies. There is for instance a question of using multicore processors which might go into the high performance computing market while WLCG prefers usage of commodity hardware.

The basic operations like distributed data management, the high data throughput and the remote job submission can probably be more cloud-like. There is a great interest among people to use commercial Clouds resources, especially when the experiments see their resources

Grid Computing in High Energy Physics Experiments 215

So, can we use Amazon or Google to do processing of data from LHC? The point is, one cannot be sure what the level of services will be and what the IN/OUT bandwidth will be. This can in principle be negotiated with these companies and may bring some level of agreement. That

As argued in [70], "Cloud Computing not only overlaps with Grid Computing, it is indeed evolved out of Grid Computing and relies on Grid Computing as its backbone and infrastructure support. The evolution has been a result of a shift in focus from an infrastructure that delivers storage and compute resources (in the case of Grids) to one that is economy based ....". Both the Grids and the Clouds communities are facing the same problems like the need to operate large facilities and to develop methods by which users/consumers

There exist a number of projects looking into and developing Cloud to Grid interfaces with the idea that Grid and Cloud Computing serve different use cases and can work together improving Distributed Computing Infrastructures (see, e.g., [71]). Also CERN is involved in

With the WLCG resources becoming used up to their limits, using commercial Clouds to process the LHC data is a strategy that should be assessed. Several of the LHC experiments have done tests whether they can use commercial Clouds. But today, the cost is rather high. Also, there are issues like whether academic data can be shipped through academic networks

Nevertheless the strategy towards deployment over the WLCG resources Cloud interfaces, managed with high level of virtualization, is under evaluation. Some level of collaboration with industry would provide the understanding how to deploy this properly and what would be the cost. The Cloud and Grid interfaces can be deployed in parallel or on top of each other. This development might also give a way to evolve into a more standardized infrastructure

A testbed of such an architecture is the CERN LXCloud [72] pilot cluster. Implementation at CERN allows to present a Cloud interface or to access other public or commercial Clouds. This is happening with no change to any of the existing Grid services. Another interesting example is the development of a comprehensive OpenSource IaaS (Infrastructure as a Service) Cloud distribution within the StratusLab project [71], see Figure 26. Anyone can take the code and deploy it on his site and have IaaS Cloud running on his site. The project is focused on deploying Grid services on top of this Cloud, 1) to be a service to existing European Grid infrastructures and to enable these people to use Cloud-operated resources and 2) because the developers consider the Grid services very complex and making sure they run safely on this

Cloud should guarantee that also other applications will run without problems.

discover, request and use resources provided by the centralized facilities.

this activity together with other international laboratories in Europe.

to a commercial provider or how to make sure what happens to this data.

and allow to make a transparent use of commercial Clouds.

becoming full.

in principle is doable.

**7.6 Grids and Clouds**
