Today, with concepts such as cloud computing and cloud services emerging one after another, databases placed in the cloud seem to be nothing new. Here we will introduce Amazon's newly launched Amazon RDS, which is the MySQL database as a cloud service. 51CTO recommends "MySQL Database Introduction and Mastery Tutorial"
Amazon recently added a new MySQL database to their Amazon Web Services (AWS) platform called Amazon Relational Database Service (RDS), which works like the traditional MySQL system. Before RDS, customers had several options for AWS database services:
Customer-provided database service running on Amazon Machine Image (AMI)
SimpleDB service owned by Amazon Web Services
SimpleDB is a simple data store that lacks the full functionality of a full-fledged relational database management system (RDBMS), but provides a scalable key-value store. The database service provided by the customer is not much different from the traditional data center environment. The customer's own employees are responsible for managing the database application, including configuration, performance tuning, capacity management, version upgrades, patching and data backup, etc. You can control it in the same way using interactive tools that connect to a traditional MySQL database.
Amazon RDS allows customer employees to reduce many MySQL operation and maintenance tasks. With it, the scalability and performance monitoring of database computing resources do not require human intervention. Database software is usually patched and backed up by the service provider, and the retention period is defined by the customer. Scalability comes from what AWS calls "instance classes," of which there are five. You can gradually increase from an ordinary virtual CPU core and 1.7G of memory (called a "small database instance") to a "very large database instance", which is 68G of memory and 8 virtual CPU cores, and the backup storage After the active database data is 100% occupied, additional storage space will be charged. And the data exists in a different Availability Zone than where the instance is. This concept is similar to the off-site data protection concept of the traditional data security model.
The service benefits from flexibility, with AWS defining a 4-hour weekly maintenance window. This maintenance window can be used to patch application software and back up data. Customers cannot opt out of the patching process. But they can specify when the maintenance window occurs during the week. During a maintenance window, a database instance is taken offline for a specific period of time. Amazon states that "Only rare occasions will patching require more than part of your maintenance window, and even if it does occur, it will only be for security or durability-related patches."
This means customers must anticipate and plan for such a weekly instance offline event. Even if the provider says it's unlikely to run out of four hours, customers expect a worst-case scenario of four hours of instance offline time per week. For customers who can accept a database instance being unavailable for a relatively short period of time, a planned shutdown with minimal possible impact may be acceptable. But some customers don't have that freedom of choice. They must keep the service available 24x7, even during weekly maintenance windows. Database replication technology is often used to achieve high availability in traditional database deployments. Can replication technology also be used in RDS, so that customers can specify different maintenance timings for different database instances? For example, are the following situations possible?
◆2 or more instances running in master-slave mode?
◆2 instances running in master-master mode?
◆2 or more instances running in cluster mode?
There is no clear answer yet. In the "Upcoming New Features" section of the RDS service details page, Amazon anticipates that the options for data replication availability will be:
Provides high availability - For developers and businesses who want the flexibility beyond Amazon RDS's automated backups, there will be no additional charge for this. With high availability support, they can easily and cost-effectively synchronously replicate database instances across multiple Availability Zones to prevent failures caused by a single storage.
It looks like this will solve availability issues at the expense of multiple availability zones. Traditional technologies to solve availability problems, such as master-slave and master-master models, cannot play a role at this point.