On Robustness and Countermeasures of Reliable Server Pooling Systems Against Denial of Service Attacks

Abstract

Abstract. The Reliable Server Pooling (RSerPool) architecture is the IETF's novel approach to standardize a light-weight protocol framework for server redundancy and session failover. It combines ideas from different research areas into a single, resource-efficient and unified architecture. While there have already been a number of contributions on the performance of RSerPool for its main tasks -pool management, load distribution and failover handling -the robustness of the protocol framework has not yet been evaluated against intentional attacks.The first goal of this paper is to provide a robustness analysis. In particular, we would like to outline the attack bandwidth necessary for a significant impact on the service. Furthermore, we present and evaluate our countermeasure approach to significantly reduce the impact of attacks.

On the application of anomaly detection in Reliable Server Pooling systems for improved robustness against denial of service attacks

“…But due to the restriction of RSerPool to a single administrative domain, a protection of the small number of PRs is assumed to be feasible [7]. Instead, the most likely attack targets are the PEs and PUs.…”

“…An attacker masquerading as a PU has two possibilities to degrade a pool's service [7], [35]: First, it can flood PRs with handle resolution requests. Since the handlespace management -and therefore the server selection procedure -can be realized very efficiently (as shown in [1], [21], [36]), such an attack requires a large attack bandwidth to cause problems.…”

“…4. The Impact of a Compromised Pool Element as possible [7], [35]: each registration request sent to a PR simply has to contain another PE ID (randomly chosen), which makes it appear as registration of a new PE. Obviously, using an adaptive pool policy, the attacker will also set the policy information for the fake PE to be selected as often as possible (e.g.…”

“…While there have already been a number of publications on the performance of RSerPool for load balancing [3]- [5] and server failure handling [6], there has been very little research on security and robustness. Until now, only some simple thresholds to avoid flooding the pool management with misinformation have been analysed in [7]. The underlying transport protocol SCTP 2 already provides protection against blind flooding attacks [9] and the Internet Draft [10] of RSerPool mandatorily requires applying mechanisms like TLS [11] or IPSEC [12] in order to ensure authenticity, integrity and confidentiality.…”