Solid-state drives (SSDs) have been widely deployed in desktops and datacenters. However, SSDs suffer from bit errors, and the bit error rate is timedependent since it increases as an SSD wears down. Traditional storage systemsmainly use parity-based RAID to provide reliability guarantees by stripingredundancy across multiple devices, but the effectiveness of RAID in SSDsremains debatable as parity updates aggravate the wearing and bit error ratesof SSDs. In particular, an open problem is that how different paritydistributions over multiple devices, such as the even distribution suggested byconventional wisdom, or uneven distributions proposed in recent RAID schemesfor SSDs, may influence the reliability of an SSD RAID array. To address thisfundamental problem, we propose the first analytical model to quantify thereliability dynamics of an SSD RAID array. Specifically, we develop a"non-homogeneous" continuous time Markov chain model, and derive the transientreliability solution. We validate our model via trace-driven simulations andconduct numerical analysis to provide insights into the reliability dynamics ofSSD RAID arrays under different parity distributions and subject to differentbit error rates and array configurations. Designers can use our model to decidethe appropriate parity distribution based on their reliability requirements.
展开▼