AWS re:Invent 2014 | (APP310) Scheduling Using Apache Mesos in the Cloud
1 668
92.7
Amazon Web Services774 тыс
Следующее
Популярные
Опубликовано 17 ноября 2014, 22:55
How can you reliably schedule tasks in an unreliable, autoscaling cloud environment? This presentation talks about the design of our Fenzo scheduler, built on Apache Mesos, that serves as the core of our stream-processing platform, Mantis, designed for real-time insights. We focus on the following aspects of the scheduler:
- Resource granularity
- Fault tolerance
- Bin packing, task affinity, stream locality
- Autoscaling of the cluster and of individual service jobs
- Constraints (hard and soft) for individual tasks such as zone balancing, unique, and exclusive instances
This talk also includes detailed information on a holistic approach to scheduling in a distributed, autoscaling environment to achieve both speed and advanced scheduling optimizations.
- Resource granularity
- Fault tolerance
- Bin packing, task affinity, stream locality
- Autoscaling of the cluster and of individual service jobs
- Constraints (hard and soft) for individual tasks such as zone balancing, unique, and exclusive instances
This talk also includes detailed information on a holistic approach to scheduling in a distributed, autoscaling environment to achieve both speed and advanced scheduling optimizations.
Свежие видео
6 дней – 5755:33
9 дней – 2 3750:19
10 дней – 2 4211:10
10 дней – 13 2084:52
10 дней – 5 9674:56
16 дней – 1260:31Случайные видео
333 дня – 2809:22
26.10.23 – 181 92816:34
19.04.23 – 22 92815:19
06.02.22 – 3 24911:16
06.08.21 – 33 1670:14
20.08.11 – 37 0013:31
1 день – 3 5595:33
2 дня – 556 5828:53
2 дня – 1 225 57525:24
4 дня – 19 11624:03
9 дней – 7985:51
12 дней – 4680:20
15 дней – 5 907 9401:08
4 дня – 7250:45