PORTLAND, Ore. — IBM is limbering up its cloud storage. Already committing $1.2 billion to establish for-rent datacenters in every country that needs them, today IBM announced a complete hardware/software/storage capability for cloud-based and hybrid users at a claimed 90 percent savings. Key is its software-defined storage-as-a-service, the newest element of IBM's software-defined storage portfolio. Called "Elastic Storage on Cloud," the new capability offered by SoftLayer (which IBM acquired last year) is an outgrowth of the Watson data management system, which beat the world champion humans at TV game show Jeopardy.
"With Elastic Storage on Cloud, we are addressing a new generation of workloads -- cloud analytics and mobile social [trending] -- which are heavily reliant on unstructured data, files, and objects," Bernie Spang, vice president of strategy for software-defined environment told EE Times. "You can have data that needs to be high performance in a flash cache on the server for high performance while moving seldom-used data that does not need low-latency access off to tape -- and it's not archived, it's active. It's just policy-based transparent movement of data to tape for up to 90 percent cost savings."
IBM's new software-defined storage-as-a-service -- Elastic Storage on Cloud -- allows its SoftLayer subsidiary to provide organizations with data management and storage in the cloud as well as seamless integration with their on-premises infrastructures.
Many of the modern datasets used today, according to IBM, deal with file sizes that exceed the limit of standard file formats, giving IBM's elastic storage concept -- running on bare metal with no virtualization (to provide maximum security) -- an advantage in capability as well as cost.
Louise Westoby, product manager for Elastic Storage on Cloud, told EE Times:
Clients today need scalable storage solutions -- NSF's 100 terabyte limit was really limiting cloud usage for these types of workloads -- particularly in scientific and technical computing, which is dealing with petabytes of data. They also needed a better way to manage data that was [both] on-premises and in the cloud.
Our solution is to offer elastic storage as a service on the software cloud for our clients -- which is taking the cloud one step further than just making the software available, but instead offers a complete package including the software infrastructure as well as full support to help the client get their data up and running in the cloud.
Also many technical and scientific programs today only occasionally need the supercomputer-calibre performance offered by cloud computing centers, making it a chore to upload petabytes of data just for a few runs of the same software they are mostly running locally on their own servers. IBM claims its cloud-based solution makes it easier for users to run in "hybrid" modes where some datasets are processed locally while others are run on the cloud.
"One of our goals is to make this experience as easy as possible from the user's standpoint," Westoby told us:
Making it easier by having a hybrid-based cloud environment, by having the software infrastructure available locally as well as available in the cloud-based environment, and minimizing the administrative burden and the cost burden, yields a complete solution with a ready-to-run software, storage, and service environment in the cloud.
Users can choose between IBM's Platform Symphony or Platform Load Sharing Facility (LSF) workload management software operating over InfiniBand networks on dedicated bare-metal servers located at the user's choice of 40 datacenters IBM promises by the end of 2014.
— R. Colin Johnson, Advanced Technology Editor, EE Times