February 22, 2013 | Written by: Steve Uniack
Share this post:
These days we hear a lot about public storage clouds and, more specifically, consumer-oriented storage services like Dropbox, Google Drive or Apple iCloud. However, corporate IT environments more often than not have a number of requirements that these public cloud services can’t always meet in terms of enterprise security (including legal and privacy), performance (primarily as a result of network latency and bandwidth) and service level agreements (SLAs). As we have seen with private compute cloud infrastructures, storage represents another significant opportunity to use the cloud computing model for private (internal) infrastructure optimization.
Within IBM’s own internal IT operating environment, satisfying the infrastructure needs for over 400,000 employees in just about every corner of the globe while concurrently supporting the storage demands of 4,000-plus business applications requires ongoing infrastructure optimization. We have been using private cloud-based solutions to meet these storage demands while continuing to deliver on our enterprise-level requirements. Like most IT organizations, within IBM we use two primary types of storage, file (network attached) and block (direct or fiber attached). I’ll review our block storage environment in a future post and focus here on our file-based private cloud storage solution.
Our private file storage cloud was originally conceived long before the widely accepted definition of “cloud computing” existed, well before cloud was even considered cool. The IBM Global Storage Architecture File Cloud service, or GSA File Cloud, was initially developed to replace existing distributed file system environments as well as local branch office and data center OS/2 and Windows-based file storage servers in the early 2000s. Over the years, GSA File Cloud has evolved into a full-fledged self-service-based file storage private cloud environment, designed to provide a consistent set of file system services across 37 worldwide locations.
GSA File Cloud provides both sharable file storage for IBM employees to use for project activities and a highly available and scalable network-attached storage (NAS) environment for many of our key business applications. IBMers use this cloud service to share files with project teams and store secondary copies of key files located on their laptops/desktops. An archive function allows users to identify specific folders that can be stored directly to lower-cost tape for up to 25 years. GSA File Cloud also has a built-in HTTP service, which allows users to easily host internal static websites accessible anywhere within the IBM intranet. Internal chargeback is pay per use and billed directly to the owning employee department on a monthly basis with reduced rates for archive and temporary storage. All of these capabilities are entirely self-service and automated using an internally accessible web-based portal.
With high availability built into its design, this private storage cloud service is also used by our most critical business applications to store and serve static web content and as storage for application log files. It is also used as the primary storage for our internal social business computing platform based on IBM Connections.
File system services traditionally require high bandwidth and low latency network connectivity; having a single global deployment would not meet the user or business application requirements. The GSA File Cloud is designed around a standard architecture deployment using common hardware and software components. As such, each physical location or “cell” consists of a set of IBM Power servers, various local or SAN attached storage options, depending on a specific location capacity requirements and IBM tape storage systems for backup and archive functions. On the software side, IBM General Parallel File System provides the highly available clustered file systems. Load balancing using IBM WebSphere, directory and web services makes up the remainder of the core software architecture components.
A set of simple web-based and command line tools are available to provide users, larger project teams and application administrators the ability to manage their storage space and permissions. GSA File Cloud also supports a number of protocols, including HTTP, FTP, SMB, NFS, SCP and SFTP. More recently, a RESTFul API based on the Storage Network Industry Association (SNIA), Cloud Data Management Interface (CDMI), has been incorporated to allow new cloud-based storage access.
Private storage clouds, like GSA File Cloud, offer significant savings opportunities for enterprises that need to consolidate existing storage infrastructure and provide self-service cloud-based services to their internal application portfolio and support the evolving storage requirements of their employees. Within IBM, we have evolved our private file storage cloud to meet these demands as well as the global nature of our environment and customers. However, the need for private storage clouds is not limited to IT organizations as large as ours. Certainly organizations of all sizes can take this same approach using vendor-provided, integrated storage solutions.
In a subsequent post, I’ll give an overview of our work in the block storage area and show where we are applying storage automation management and tiering to achieve significant storage cost savings and improve overall utilization.