Performance
-
Murat and Aleksey Read Papers: “Rethinking the Cost of Distributed Caches for Datacenter Services”
Murat and Aleksey read “Rethinking the Cost of Distributed Caches for Datacenter Services.“ This paper argues that distributed caches can save money by reducing CPU costs despite using substantially more of the costly DRAM. The authors claim up to 4X savings in synthetic and open-source workloads. The paper also calls for richer caching semantics to…
-
On Metastable Failures and Interactions Between Systems
I’ve written about metastable failures before. The topic has been picked up by a few different teams since the, all analyzing metastable failures more, while I apparently has been slacking off… Anyway, Metastable failures are self-sustaining performance failures that arise in systems due to a positive feedback loop triggered by an initial problem. This positive…
-
HoliPaxos: Towards More Predictable Performance in State Machine Replication
I will be presenting our new paper, “HoliPaxos: Towards More Predictable Performance in State Machine Replication,” at the VLDB’25. Feel free to ping me if you are there and want to chat! This paper explores several orthogonal optimizations to the classical MultiPaxos state machine replication protocol to improve its performance stability in the presence of…
-
Paper #192. OLTP Through the Looking Glass 16 Years Later: Communication is the New Bottleneck
Last week we read “OLTP Through the Looking Glass 16 Years Later: Communication is the New Bottleneck” CIDR’25 paper by Xinjing Zhou, Viktor Leis, Xiangyao Yu, Michael Stonebraker. This paper revisits the original “OLTP Through the Looking Glass, and What We Found There” paper and examines the bottlenecks in modern OLTP databases. The new paper…
-
Paper #191: Occam’s Razor for Distributed Protocols
We have been doing a Zoom distributed systems paper reading group for 5 years and have covered around 190 papers. This semester, we should reach the milestone of 200 papers. Over the years, my commitment to the group has varied — at some point, I was writing paper reviews, and more recently, I’ve had less…
-
Pile of Eternal Rejections: The Cost of Garbage Collection for State Machine Replication
I have a “pile” of papers that continuously get rejected from any conference. All these papers, according to the reviews, “lack novelty,” and therefore are deemed “not interesting” by the reviewing experts. There are some things in common in these papers — they are either observational or rely on old and proven techniques to solve a problem or improve a system/algorithm. Jokingly, I call this set of papers the “pile of…
-
Reading Group 151. Towards Modern Development of Cloud ApplicationsReading Group 151.
We kicked off the winter term set of papers in the reading group with the “Towards Modern Development of Cloud Applications” HotOS’23 paper. The paper proposes a different approach to designing distributed applications by replacing the microservice architecture style with something more fluid. The paper argues that splitting applications into microservices from the get-go can…
-
Reading Group. Chardonnay: Fast and General Datacenter Transactions for On-Disk Databases
Last week, we looked at the “Chardonnay: Fast and General Datacenter Transactions for On-Disk Databases” OSDI’23 paper by Tamer Eldeeb, Xincheng Xie, Philip A. Bernstein, Asaf Cidon, Junfeng Yang. The paper presents a transactional database built on the assumption of having a very fast two-phase commit protocol. Coordination, like a two-phase commit (2PC), usually has…
-
Cloudy Forecast: How Predictable is Communication Latency in the Cloud?
Many, if not all, practical distributed systems rely on partial synchrony in one way or another, be it a failure detection, a lease mechanism, or some optimization that takes advantage of synchrony to avoid doing a bunch of extra work. These partial synchrony approaches need to know some crucial parameters about their world to estimate…
-
Reading Group. Rabia: Simplifying State-Machine Replication Through Randomization
We covered yet another state machine replication (SMR) paper in our reading group: “Rabia: Simplifying State-Machine Replication Through Randomization” by Haochen Pan, Jesse Tuglu, Neo Zhou, Tianshu Wang, Yicheng Shen, Xiong Zheng, Joseph Tassarotti, Lewis Tseng, Roberto Palmieri. This paper appeared at SOSP’21. A traditional SMR approach, based on Raft or Multi-Paxos protocols, involves a…
Search
Recent Posts
- Murat and Aleksey Read Papers: “Rethinking the Cost of Distributed Caches for Datacenter Services”
- On Metastable Failures and Interactions Between Systems
- Murat and Aleksey Read Papers: “Barbarians at the Gate: How AI is Upending Systems Research”
- Academic Chat with Murat and Aleksey: 5 Cs of the Invisible Curriculum.
- HoliPaxos: Towards More Predictable Performance in State Machine Replication
Categories
- One Page Summary (11)
- Other Thoughts (14)
- Paper Review and Summary (16)
- Pile of Eternal Rejections (2)
- Playing Around (14)
- Reading Group (104)
- RG Special Session (4)
- Teaching (2)