<?xml version="1.0" encoding="utf-8" standalone="yes"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>availability on Eric Hotinger</title>
    <link>https://ehotinger.com/categories/availability/</link>
    <description>Recent content in availability on Eric Hotinger</description>
    <generator>Hugo -- gohugo.io</generator>
    <language>en-US</language>
    <lastBuildDate>Fri, 18 Apr 2025 10:06:17 -0700</lastBuildDate>
    <atom:link href="https://ehotinger.com/categories/availability/index.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>Amazon ECS Set Default Log Driver Blocking Mode</title>
      <link>https://ehotinger.com/blog/amazon-ecs-set-default-log-driver-blocking-mode/</link>
      <pubDate>Fri, 18 Apr 2025 10:06:17 -0700</pubDate>
      <guid>https://ehotinger.com/blog/amazon-ecs-set-default-log-driver-blocking-mode/</guid>
      <description>Amazon ECS Set Default Log Driver Blocking Mode As of April 17th, 2025 in Amazon Elastic Container Service (ECS), you can set the defaultLogDriverMode for your tasks at an account level using the PutAccountSettingDefault API.&#xA;What is the problem with logs? Docker has a logging daemon implementation which will read STDOUT and STDERR of a container and convert logs into an in-memory ring buffer (see code here). Those messages get sent to a log driver, such as the AWS log driver, which can then upload them to a service such as CloudWatch.</description>
    </item>
    <item>
      <title>Operating in Grey Modes</title>
      <link>https://ehotinger.com/blog/operating-in-grey-modes/</link>
      <pubDate>Thu, 19 Jan 2023 13:26:55 -0800</pubDate>
      <guid>https://ehotinger.com/blog/operating-in-grey-modes/</guid>
      <description>Grey Modes In complex distributed systems, hardware failure is inevitable. The ability to operate in a grey mode (a mode in which a system is partially working) is an important and deliberate architectural choice to consider. Ask yourself the following question: would customers be more upset to have a completely nonfunctional system, or have a subset of the functionality working? For example, if you are working on a storage layer, perhaps it would be better to serve reads even though you can&amp;rsquo;t serve writes at the same time.</description>
    </item>
    <item>
      <title>Availability Numbers Every Programmer Should Know</title>
      <link>https://ehotinger.com/blog/availability-numbers-every-programmer-should-know/</link>
      <pubDate>Sun, 08 Jan 2023 07:22:14 -0800</pubDate>
      <guid>https://ehotinger.com/blog/availability-numbers-every-programmer-should-know/</guid>
      <description>Availability Numbers Every Programmer Should Know Availability refers to the amount of continuous uptime of a service. This is measured in a percentage, where 100% means a service has zero downtime and 0% means the service is never operational.&#xA;A service level agreement (SLA) is a commitment from a provider to a customer. When SLAs are mentioned for availability, providers will be stating their commitment to the uptime of the system, after which you are usually entitled to some form of compensation.</description>
    </item>
  </channel>
</rss>
