Following the huge success of being able to provision a consistent, user-requested I/O rate forDynamoDBandElastic Block Store(EBS), the AWS Database Services team has now released Provisioned IOPS, a new high performance storage option for the下面的n Relational Database Service(Amazon RDS). Customers can provision up to 10,000 IOPS (input/output operations per second) per database instance to help ensure that their databases can run the most stringent workloads with rock solid, consistent performance.

High Performance I/O

下面的n RDS Provisioned IOPS is intended for applications that need predictable performance and have database workloads that generate largely random I/O. Amazon RDS Provisioned IOPS is ideal for mission-critical online transaction processing (OLTP) workloads that require a high performance storage option with consistent IOPS, within a narrow band of tolerance.

As with all AWS services, if you are in doubt about whether your application can benefit from Provisioned IOPS, it is very easy to try it out and decide for yourself, as there are no upfront costs.

By the way, for mission critical OLTP workloads, you should also consider adding the Amazon RDS Multi-AZ option to improve availability.



With the ability to provision IOPS to deliver consistent performance from a database, customers should be able to significantly reduce the complexities in their application because the database is no longer the bottleneck in their architectures. As a consequence, these simpler architectures are more cost-effective and easier to make reliable because there are fewer moving parts. For example, with the consistency you can get with Amazon RDS Provisioned IOPS, you may be able to dial up your IOPS and consolidate some of the write shards or read replicas you may be using in your architecture today.

Last week I reada great example为EBS提供EOPS对客户的架构进行了直接影响。解析,在AWS上运行的非常酷的移动开发平台,切换到必威体育精装版app官网使用配置的I / O用于EBS运行其数据库。在一周内,最终用户的延迟被削减了一半,并且在其架构中不再存在性能波动。此外,不再需要他们过去必须跳过预热数据库的箍。提供I / O的能力使它们更有表现,较少的努力。阅读更多信息他们的博客.

Sang Chi ofFlipboardreported that because of the high I/O performance and consistency they are seeing with Amazon RDS Provisioned IOPS, they expect to scale their database applications to tens of thousands of IOPS. These are great examples of how many of our AWS customers are successful in making their database achieve higher performance while having to do less work in managing their databases.

Provisioned IOPS storage in RDS

The Provisioned IOPS storage option for RDS will be rolled out in two phases. In the first phase, IOPS can be provisioned for use with new database instances in increments of 1,000 IOPS up to a maximum of 10,000 IOPS if you are using the MySQL or Oracle database engines. If you are using the SQL Server database engine, you can provision from 1,000 IOPS to 7,000 IOPS. With all databases engines you configure corresponding storage from 100GB to 1TB.

In the second phase, we plan to provide an automated mechanism through the AWS Console and the RDS APIs, that will allow you to migrate existing databases running on Standard RDS storage to Provisioned IOPS storage, if you are using the MySQL or Oracle database engines. If you want to migrate an existing RDS database instance to Provisioned IOPS storage immediately, you can export the data from your existing database instance and import into a new database instance equipped with Provisioned IOPS storage.

The performance you get out of your databases will depend on the type of database you are using and your workloads. The Amazon RDS Provisioned IOPS storage option is available for all engines supported by Amazon RDS: MySQL, Oracle, and Microsoft SQL Server.

Amazon RDS提供IOPS立即可用于美国东(N.弗吉尼亚州),美国西部(N.California)和Eu West(爱尔兰)地区的新数据库实例。该团队计划在未来几个月内在其他AWS地区启动。

If you want to learn more about Amazon RDS Provisioned IOPS storage for Amazon Relational Database Service visit the下面的n RDS detail page,下面的n RDS User Guideand theProvisioned IOPS Technical FAQ.You can also learn how customers are already taking advantage of Amazon RDS Provisioned IOPS by reading theblog post on the AWS developer blog.


blog comments powered byDisqus