| SSD has been playing a significantly important role in caching systems due to high readwrite performance and cost-effective.Since cache space is much smaller than that of the backend storage by one order of magnitude or even more,write density(writes per unit time and space)of SSD cache is therefore much higher than that of HDD storage,which brings about great challenges to SSD's lifetime.Meanwhile,under social network workloads,quite a lot writes on SSD are unnecessary,e.g.,in a famous company's photo caching,there is 61% of total photos are just accessed once whereas they are still swapped in and out of the cache.Since these images are only accessed once,they will not be accessed again after entering the cache.It is useless to write such images to the cache.Therefore,if we can predict this kind of photos proactively and prevent them from entering the cache,we can eliminate unnecessary SSD cache writes and improve cache space utilization.To cope with the challenge,we put forward a "one-time-access criteria" that is applied to cache space,and further propose a "one-time-access-exclusion" policy.Based on that,we design a prediction-based classifier to facilitate the policy.Unlike the state-of-the-art historybased predictions,our prediction is non-history-oriented,the prediction here needs to be judged on the first access,and there is no history information,which is challenging to achieve a good prediction accuracy.To address this issue,we integrate a decision tree into the classifier,extract social-related information as classifying features,and apply history table,costsensitive learning and daily update model to improve classification precision.Due to these techniques,we attain prediction accuracy of over 80%.Experimental results show that "onetime-access-exclusion" approach makes caching performance outstanding in most aspects,taking LRU for instance,hit rate is improved by 4.4%,cache writes are decreased by 56.8%,and the average access latency is dropped by 5.5%. |
PDF Full Text Request