Citation Context

...ooperative. One way to address this problem is to retrofit support for performance isolation to the operating system. This has been demonstrated to a greater or lesser degree with resource containers =-=[3]-=-, Linux/RK [32], QLinux [40] and SILK [4]. One difficulty with such approaches is ensuring that all resource usage is accounted to the correct process — consider, for example, the complex interactions...

Citation Context

...e to product groups, so that individual groups do not need to maintain their own clusters. As our service clusters scale, we will need to deal with more resource-sharing issues within Bigtable itself =-=[3, 5]-=-. Finally, we have found that there are significant advantages to building our own storage solution at Google. We have gotten a substantial amount of flexibility from designing our own data model for ...

Citation Context

...fortunately, mechanisms for dealing with denial-of-service have not advanced at the same pace. Most work in this area has focused on tolerating attacks by mitigating their effects on the victim [40], =-=[2]-=-, [27], [30], [9]. This approach can provide an effective stopgap measure, but does not eliminate the problem, nor does it discourage attackers. The other option, and the focus of this paper, is to tr...

Citation Context

...l request stream to implement such a policy; all it knows is that the thread pool is saturated, and must arbitrarily reject work without knowledge of the source of the bottleneck. Resource containers =-=[7]-=- and the concept of paths from the Scout operating system [41, 49] are two techniques that can be used to bound the resource usage of tasks in a server. These mechanisms apply vertical resource manage...

Citation Context

...orized as being focused on tolerance, diagnosis and localization. The first category is composed of both approaches for mitigating the impact of specific attacks [4, 16] and general system mechanisms =-=[25, 1]-=- for controlling resource usage on the victim machine. Usually such solutions involve a quick triage on data packets so minimal work is spent on the attacker’s requests and the victim can tolerate mor...

Citation Context

...gy" to refer to NLANR measurement data [16], which represent only a part of the actual Internet AS graph. have focused on detecting DoS attacks and mitigating their detrimental impact upon the vi=-=ctim [1, 13, 24, 26]-=-. This approach does not eliminate the problem, nor does it necessarily deter potential attackers. A number of recent works have studied source identification (also called IP traceback [23]) which spa...

Citation Context

...or research efforts. Much work focused on CPU scheduling and resource allocation such as Hierarchical scheduling [30], processor capacity reserves [42], CPU reservations [33], and resource containers =-=[15]-=-. In contrast, we develop a resource management architecture in middleware which can run on top of any standard operating system, thereby creating a more portable solution. Our work [5, 6, 3, 40] diff...

Citation Context

... fairness between vservers; nor can it provide guarantees. PlanetLab’s CPU scheduler uses a proportional sharing (PS) scheduling policy to fairly share the CPU. It incorporates the resource containe=-=r [3] a-=-bstraction and maps each vserver onto a resource container that possesses some number of shares. Individual processes spawned by the vserver are all placed within the vserver’s resource container. T...