Amazon Spins Up DynamoDB Database as a Service
Jan 19, 2012 11:43 AM PT
Amazon.com has launched its DynamoDB fully managed NoSQL database service, targeting businesses that need to deal with large amounts of data.
The service will use the company's traditional pay-as-you-go model, and Amazon will take care of the administrative side of things, such as hardware provisioning and configuration, replication and partitioning.
DynamoDB is based on provisioned throughput, meaning that customers will state how much read/write throughput they need when they create database tables using the service.
The service is highly scalable, and scaling is done online. Customers can go from minuscule requirements such as five reads per second to even 50,000 reads, for example.
Charges will kick in when customers go beyond 100 MB of free storage, five writes per second and 10 strongly consistent reads per second, which are offered at no cost, as part of the Amazon Web Services (AWS) Free Tier.
DynamoDB and Hadoop
Amazon DynamoDB is integrated with Amazon Elastic MapReduce (EMR), which is a pay-as-you-go hosted Apache Hadoop service running on Amazon Web Service.
Hadoop from Apache is a software framework that supports data-intensive distributed applications. It lets them work with petabytes of data across thousands of nodes.
Although DynamoDB has just been launched, the concept behind it may appear familiar.
"The pay-as-you-go NoSQL plus Hadoop offering has been around for years," Justin Sheehy, chief technology officer at Basho Technologies, told TechNewsWorld.
DynamoDB stores data on solid-state drives and replicates it synchronously across multiple AWS availability zones in an AWS region to provide high availability and data durability.
The 10 reads per second offered free for starters works out to 40 million requests a month, Amazon said.
DynamoDB's integration with EMR lets customers use work with data stored across multiple locations and archive the result separately from the original database in DynamoDB.
Getting Started With DynamoDB
Customers create DynamoDB tables and tell Amazon how much read/write throughput they need. They can request changes in throughput as needed.
Newly created tables will usually be ready to use within a couple of minutes. Once they're ready, customers store their data, paying only for the storage they use. AmazonDB will provision adequate storage.
Customers can create up to 56 tables, each provisioned for 10,000 reads and 10,000 writes per second.
AWS customers will see a DynamoDB tab in their AWS Management Consoles which lets them configure DynamoDB service easily.
Amazon has updated the AWS software development kits to include support for DynamoDB.
Safety and Security
One of the issues with using cloud services is governance. Since servers are called up dynamically in the cloud as needed, questions may be raised about whether or not regulated data may have been stored on the same server as unregulated data, for example.
"AWS has received several compliance certifications for the operations of its infrastructure and its services, as well as FISMA low and moderate [certifications] for the handling of government data," Werner Vogels, chief technology officer of Amazon, told TechNewsWorld.
Further, AWS has "achieved PCI Validated Level 1 Service Provider Status," which lets customers process, store or transmit credit card data on AWS servers," Vogels said. "Protecting our customers will always be AWS's number one priority."
FISMA, the Federal Information Security Management Act, requires United States federal government agencies to develop, document and implement an IT security program.
"Amazon might not be able to handle certain kinds of data because of specific complexities or situations, but there's plenty of regulated, auditable data on Amazon today," Basho's Sheehy said.
It's Not Always All Good
Although DynamoDB imposes usage-based charges, that doesn't necessarily make it more cost-effective than creating and maintaining a database in-house.
"It's very expensive to use Amazon in bulk over time, so many companies will likely find the pay-as-you-go model most attractive during the prototyping phase to get up and running," Basho's Sheehy pointed out.