Benchmarking big data systems: A review
With the fast development of big data systems in recent years, a variety of open-source
benchmarks have been built to evaluate and compare the workloads on these systems, and …
benchmarks have been built to evaluate and compare the workloads on these systems, and …
Ernest: Efficient performance prediction for {Large-Scale} advanced analytics
Recent workload trends indicate rapid growth in the deployment of machine learning,
genomics and scientific workloads on cloud computing infrastructure. However, efficiently …
genomics and scientific workloads on cloud computing infrastructure. However, efficiently …
Tiny-tail flash: Near-perfect elimination of garbage collection tail latencies in NAND SSDs
Flash storage has become the mainstream destination for storage users. However, SSDs do
not always deliver the performance that users expect. The core culprit of flash performance …
not always deliver the performance that users expect. The core culprit of flash performance …
{FlashBlox}: Achieving Both Performance Isolation and Uniform Lifetime for Virtualized {SSDs}
A longstanding goal of SSD virtualization has been to provide performance isolation
between multiple tenants sharing the device. Virtualizing SSDs, however, has traditionally …
between multiple tenants sharing the device. Virtualizing SSDs, however, has traditionally …
The tail at store: A revelation from millions of hours of disk and {SSD} deployments
M Hao, G Soundararajan… - … USENIX Conference on …, 2016 - usenix.org
We study storage performance in over 450,000 disks and 4,000 SSDs over 87 days for an
overall total of 857 million (disk) and 7 million (SSD) drive hours. We find that storage …
overall total of 857 million (disk) and 7 million (SSD) drive hours. We find that storage …
Understanding manycore scalability of file systems
We analyze the manycore scalability of five widelydeployed file systems, namely, ext4, XFS,
btrfs, F2FS, and tmpfs, by using our open source benchmark suite, FXMARK. FXMARK …
btrfs, F2FS, and tmpfs, by using our open source benchmark suite, FXMARK. FXMARK …
The unwritten contract of solid state drives
We perform a detailed vertical analysis of application performance atop a range of modern
file systems and SSD FTLs. We formalize the" unwritten contract" that clients of SSDs should …
file systems and SSD FTLs. We formalize the" unwritten contract" that clients of SSDs should …
MittOS: Supporting millisecond tail tolerance with fast rejecting SLO-aware OS interface
MittOS provides operating system support to cut millisecond-level tail latencies for data-
parallel applications. In MittOS, we advocate a new principle that operating system should …
parallel applications. In MittOS, we advocate a new principle that operating system should …
On the performance variation in modern storage stacks
Z Cao, V Tarasov, HP Raman, D Hildebrand… - … USENIX conference on …, 2017 - usenix.org
Ensuring stable performance for storage stacks is important, especially with the growth in
popularity of hosted services where customers expect QoS guarantees. The same …
popularity of hosted services where customers expect QoS guarantees. The same …
Managing tail latency in datacenter-scale file systems under production constraints
Distributed file systems often exhibit high tail latencies, especially in large-scale datacenters
and in the presence of competing (and possibly higher priority) workloads. This paper …
and in the presence of competing (and possibly higher priority) workloads. This paper …