The amount of storage being generated everyday by enterprises is growing rapidly, with no sign of slowing down. Enterprise IT is now faced with the challenge of storing and managing this data in a world where IT budgets are stagnant or even in decline. Many organizations are turning to OpenStack Swift to drive down the cost of storage and to handle the massive influx of data. But while the flexibility of open software is appealing, its do-it-yourself (DIY) nature, and lack of enterprise-class support is not. An OpenStack project often requires more time than expected for enterprises to support long term.
What is OpenStack and OpenStack Swift?
OpenStack is a set of software tools for building and managing computing platforms for public and private clouds, backed by some of the biggest companies in software development and hosting like Rackspace, IBM and HP. Swift is the object storage project within the OpenStack community. Object storage is an alternative means to accessing data to the more traditional POSIX file system. Object storage allows more rapid access to a greater number of files (objects) than do these file systems. Swift brings this type of architecture to the OpenStack initiative, at massive scale supporting globally concurrent access to storage.
OpenStack Swift, while providing the base tools needed to build and integrate an object storage architecture, requires a lot of ‘heavy lifting’ on the part of the application provider or the enterprise customer. This means it takes time to deploy and integrate with the existing enterprise systems and infrastructure, while keeping up with the fast pace of data growth. Swift also requires consistency and precision; one missed step in the integration process may lead to intermittent errors that could take weeks to resolve without the help of a support team dedicated to the success of the project.
What is SwiftStack?
SwiftStack powers enterprises with a software defined storage platform that delivers a massively scalable private cloud. The SwiftStack platform simplifies storage management and ensures uninterrupted scaling for multi-geographic data centers, while ensuring seamless integration with existing enterprise systems and infrastructures. By deploying SwiftStack for archiving, content delivery and disaster recovery enterprises, including HP, eBay and Pac-12 have substantially decreased overall TCO. The SwiftStack platform is comprised of an out-of-band Software Defined Controller, SwiftStack Nodes, the Filesystem Gateway, and at its core, the OpenStack Swift object storage architecture, which also powers the world’s largest public clouds.
Basically, SwiftStack completes the puzzle by including the features and capabilities that application providers and enterprise data centers need in a more turnkey, consumable package. The result is an easy to manage and cost effective object storage deployment with enterprise-class support.
While saving on storage costs is important to both application providers and enterprises, wringing the last possible cent out of those costs is typically not. Most of these users are willing to spend a little extra to get the integrated services and support they need so they can return to running their businesses. In many ways SwiftStack is to OpenStack Swift what RedHat is to Linux.
How Much Heavy Lifting is There with OpenStack Swift?
Since the major difference between OpenStack Swift and SwiftStack is the amount of effort involved, it is important to detail what that effort entails. After all, if the organization has the time and skill set to do this heavy lifting then there might be some advantage to building it themselves. The work required can be grouped into four categories;
- Automation management
- Monitoring and capacity management
- Professional services
The first category to consider is integration; how hard it will be to integrate the object storage software into the existing environment with existing user profiles. OpenStack is lacking the native ability to connect to LDAP (Lightweight Directory Access Protocol) or to Active Directory Services. While in theory this could be added to OpenStack Swift, doing so would require significant programming effort.
Part of integration includes integration with existing file systems and data access methods. OpenStack Swift does not provide any file system gateways. Instead the enterprise is faced with having to migrate all their applications to native REST or HTTP access. SwiftStack on the other hand provides a file system gateway supporting both CIFS and NFS access. This eliminates the need for additional add on software that can further complicate as well as increase the cost of the OpenStack Swift solution.
The second area to consider is the level of automation, both for deployment and for ongoing operations. OpenStack Swift does provide a package-based implementation, but it does not provide the ability to tune performance on a per-use-case basis. Nor does it automate the process of system, disk or cluster management. The result is that it can take 6 to 12 months with several engineers dedicated to the task to build a manageable and scalable system. SwiftStack, on the other hand, claims to be able to deploy a system of this scale in less than one month, with just one engineer.
Organizations use configuration management tools or write their own scripts to automate OpenStack Swift. The problem is that the scripts are too unique to the environment; they don’t enjoy use across thousands of installations. Custom scripts will solve the problem at hand but at what cost? They’re also difficult for more than one administrator to manage and support. Custom scripts provide job security for the script writer, but if that person is unavailable the organization is left exposed.
SwiftStack systems are pre-tested and validated across SwiftStack’s hardware partners, providing a turnkey solution. SwiftStack also has automation built into its software and makes it accessible from a management GUI. This means that its automation code is available without programming and that it’s hardened across thousands of sites. The organization can change storage management responsibilities as needed.
Managing storage in OpenStack Swift is also challenging. Basic management structures are part of the core code, but advanced storage management is not. Performing functions like system upgrades and cluster management need manual, high-risk administrator interaction. There is also the potential of downtime if an administrator performs a step incorrectly or forgets it all together.
A final area of concern and of utmost importance for application providers and enterprises, where storage infrastructure is critical, is capacity management. Functions like adding capacity to a live cluster in OpenStack Swift, or safely removing a failing node or drive are all management challenges. There is also a limited number of metrics available to the administrator.
By comparison, SwiftStack has over 500 metrics per node that it analyzes and reports on. And execution of the above functions is all done through the provided GUI as well as the reporting of their progress.
Why Build it Yourself?
For most organizations, however, this is not the case. Storage is important but does not need to be a competitive differentiator. Of course, saving money is also important, but these organizations are willing to strike a balance. They achieve that balance by fast implementation, low operational costs of a more turnkey solution and reducing the risk and complexity of having to manage the entire lifecycle of a Swift cluster. SwiftStack is an excellent example of a software application providing that balance.
Sponsored by SwiftStack