July 2013 Summaries
3 posts from Redis
Filter
Month:
Year:
Post Summaries
Back to Blog
The text discusses the evolution of cloud service providers and their networks over time. It highlights that while private IP addresses were initially believed to be preferable for in-system communications, experience has shown otherwise. In particular, it explains how AWS's private addresses can sometimes perform worse than public addresses due to extra protection provided by software-based firewalls. The author recommends PaaS users to avoid using the private interface and instead use the public one, despite its slightly higher latency, for more consistent performance.
Jul 31, 2013
734 words in the original blog post.
Redis' replication mechanism is a simple yet effective tool for increasing availability and scaling out Redis setups by performing reads against read-only slaves. However, in certain rare scenarios where dataset size grows significantly, activation of replication can prove to be a challenging task due to various limitations such as latency due to fork, physical limits of file transfer, and time constraints. These limitations can lead to prolonged synchronization times, causing the slave to never complete synchronization and replication to start. To address this issue, Redis provides a client-throttling mechanism, which introduces delays to master server responses to application clients' requests, providing enough "breathing room" for the slave to finish transferring the file and replaying updates log before the latter runs out of space. The throttling mechanism is configured using the `slave-output-buffer-throttling` configuration variable, which allows users to set a low, high, rate, and delay threshold for optimal performance.
Jul 16, 2013
1,345 words in the original blog post.
Benchmarking is a crucial practice in measuring system performance to identify its limits. Our service uses benchmarking for both regression testing and optimizing database performance. We've developed our own benchmarking tool, memtier_benchmark, which provides customizable traffic patterns against Memcached and Redis instances. The tool offers robust reporting capabilities and can be used extensively for development and sharing knowledge with the community. Its source code is released under GPLv2 licensing, allowing users to download it from our GitHub account. Memtier_benchmark offers various features such as connection and general options, test options, object options, imported data options, key options, and more, making it a convenient and easy-to-use command-line interface for benchmarking.
Jul 02, 2013
511 words in the original blog post.