Amazon Web Services' Simple Storage Service now offers replication of data across regions rather than just across...
availability zones, which could mean simpler, cheaper geographic redundancy for disaster recovery.
The Simple Storage Service (S3) replication is based on S3's existing versioning functionality and enabled through the Amazon Web Services (AWS) Management Console. To get started, users can choose the destination region and bucket and then set up an Identity and Access Management role to allow the replication utility access to S3 data. There is also the option to limit replication to a subset of objects in a given S3 bucket. Once the feature is enabled, every object uploaded to the S3 bucket is automatically replicated.
The replication comes at a cost of $0.02 per gigabyte (GB) of data transfer, and S3 storage costs for the replicated data max out at $0.03 GB. Compared to some other third-party replication products available, or the cost of setting up geographically redundant data centers without AWS' infrastructure, new cross-region S3 replication could be a boon to customers looking for a quick, cheap alternative to disaster recovery, according to cloud consultants.
"It turns something that is ordinarily complex and operationally troublesome into something completely automatic and transparent," said Jared Reimer, co-founder of Cascadeo Corp., an IT consulting firm located in Mercer Island, Wash.
Not everyone needs this level of geographic redundancy, Reimer said, but for those who do, cross-region S3 replication is welcome news.
"Consider what it would have cost to build and operate this capability consistently in a conventional IT environment," Reimer said.
With this feature, new disaster recovery strategies can be developed for companies without having to maintain third-party replication software or write scripts to automate replication of S3 objects, according to Glenn Grant, CEO of G2 Technology Group based in Boston.
"With this feature we can develop a new strategy where images and backup data can be dumped into S3 instead of replicating to [Elastic Block Store] volumes in different geographies," Grant said. "It should ultimately reduce cost … [and] we will no longer have to maintain the associated replication software or services ourselves."
S3 replication alternatives
S3's new replication capabilities aren't unique; other tools have already sprung up in the market and gained followers. And S3 replication doesn't yet match up to all of the capabilities these tools can offer, according to data storage industry analysts.
"We do think that data protection capabilities will become part of fundamental storage solutions, and less based on third-party additional software layers," said Mike Matchett, analyst with the Taneja Group, Inc., a storage analyst firm based in Hopkinton, Mass.
But there is still a long way to go to really compete with existing products which might better track items and help retrieve the most optimal version when needed.
"There is no real visibility or built-in metadata indexing or tracking when using S3 simply as a replicated storage target," Matchett said.
Some replication tools also offer built-in deduplication of data, which S3 replication doesn't offer yet.
"Any deduplication that AWS is doing with S3 data on the back end would accrue a benefit only to AWS at this point -- you'll still pay for what it looks like you are using," Matchett said. "This can result in a huge difference in cost and also in recovery time."
Third party tools which replicate data across clouds and offer data deduplication include HP's Data Protector with HP StoreOnce. There is also a free tool which can replicate object storage between multiple clouds based on the S3 APIs, called S3motion, developed by EMC. Users can also program replication into S3 using the AWS Lambda utility.