Kun Lin
Working on improving fleet efficiency.
Authored Publications
Sort By
Preview abstract
CXL is a new computer bus protocol, designed to replace PCIe. Because it has much lower latency than PCIe (hundreds of nanoseconds) and hardware support for cache coherence, CXL can provide efficient bus access to remote memory. These capabilities have opened the possibility of CXL memory pools in datacenter and cloud networks, consisting of a large memory pool, dynamically shared between multiple machines.
In this paper, we give three reasons why CXL memory pools will not help datacenter or cloud applications: cost, complexity, and limited utility. While CXL memory pools can potentially decrease RAM costs through multiplexing, they require a new cabling and switching infrastructure in parallel to Ethernet, whose cost outweighs any savings. Experimental results show that CXL is substantially higher latency than main memory, such that using
it without harming application performance will require either greater application or system complexity. Finally, as modern servers have hundreds of cores and TB of RAM, they provide great flexibility in job and VM placement, such that a good system scheduler does not strand resources or have difficulty finding servers with enough RAM.
View details
WSMeter: A Fast, Accurate, and Low-Cost Performance Evaluation for Warehouse-Scale Computers
Jaewon Lee
Changkyu Kim
Rama Govindaraju
Jangwoo Kim
Proceedings of the Twenty-Third International Conference on Architectural Support for Programming Languages and Operating Systems (2018) (to appear)
Preview abstract
A warehouse-scale computer (WSC) is a vast collection
of tightly networked computers providing modern internet
services, that is becoming increasingly popular as the most
cost-effective approach to serve users at global scale. It is
however extremely difficult to accurately measure the holistic
performance of WSC. The existing load-testing benchmarks
are tailored towards a dedicated machine model and do not
address shared infrastructure environments. Evaluating the
performance of a live shared production WSC environment
presents many challenges due to the lack of holistic performance
metrics, high evaluation costs, and potential service
disruptions they may cause. WSC providers and customers
are in need of a cost effective methodology to accurately evaluate
the holistic performance of their platforms and hosted
services.
To address these challenges, we propose WSMeter, a cost
effective framework and methodology to accurately evaluate
the holistic performance of WSC in a live production environment.
We define a new performance metric to accurately reflect
the holistic performance of a WSC running a wide variety of
unevenly distributed jobs. We propose a model to statistically
embrace the performance variances amplified by co-located
jobs, to evaluate holistic performance with minimum costs.
For validation of our approach, we analyze two real-world
use cases and show that WSMeter accurately discerns 7% and
1% performance improvements, using only 0.9% and 6.6% of
the machines in the WSC, respectively. We show through a
Cloud customer case study, where WSMeter helped quantify
the performance benefits of service software optimization with
minimal costs.
View details
