Company bypasses network-attached storage for Linux clusters

A couple of years ago, Dynamic Graphics Group might have chosen network-attached storage (NAS) and not Linux for its new storage and delivery platform. But the company's IT gurus -- Todd Moore and Brian Kohles -- quickly discovered that NAS cost twice as much, delivers less I/O throughput and has fewer expansion options than a Linux cluster.

Moore and Kohles said they would not entrust Dynamic Graphics' crown jewels -- digital photography, art, movies and other licensed media -- to a flimsy storage platform that would not scale. "We had to safeguard and yet provide continual, authorized access to these assets," said Moore, IT director for Dynamic Graphics, a global media company based in Peoria, Ill. Dynamic Graphics offers subscription-based, royalty-free art services, stock image content and creative services training. Kohles is Dynamic Graphics' Web developer and Linux expert.

In this interview, Moore and Kohles tell the story of their storage and database migration to Linux. In this installment, they explain the hassles caused by their aging Unix and Windows systems and why they chose Linux to replace them. In part two, they describe the deployment and its results.

This Content Component encountered an error

Why did you end up choosing a PolyServe Matrix Server-enabled cluster of Linux servers? The PolyServe on Linux solution combined the benefits of NAS and SAN. Each server in the cluster can simultaneously read and write to volumes of shared data on a SAN. PolyServe provided high availability to ensure system uptime for the NFS, CIFS, FTP and HTTP services running on the cluster. The NAS alternatives only provided file access via HTTP,...

NFS and CIFS, and [it] required separate dedicated backup and FTP servers.


Continue reading the second part of this interview.

FEEDBACK: Is your enterprise using a Linux-based cluster over network-attached storage? We want to hear about it.
Send your feedback to the SearchEnterpriseLinux.com news team.

Why did you choose Linux?
Linux offers such an incredible kind of price-to-performance ratio. We were very comfortable with the Linux idea and the amount of flexibility that it gave us. It just became a win-win for us.
Linux offers such an incredible kind of price-to-performance ratio. We were very comfortable with the Linux idea and the amount of flexibility that it gave us. It just became a win-win for us.
Todd Moore
IT DirectorDynamic Graphics Group
What remedies, solutions and new processes were examined?
We looked at solutions from EMC, Network Appliance, Microsoft, a lot of SAN and NAS vendors, and we looked at several Linux solutions. Among other things, we looked at two Windows-based NAS systems and a multi-headed NAS gateway with SAN storage. What was on your wish list?
We needed to be able to house all of our content on a large scalable file system. It's going to be 10-to-1 terabytes to start but should be able grow to 30 or 40 without a complete redesign of the system. We wanted to be able to deliver images via HTTP for straight downloads once transactions had taken place. We needed to be able to do backups without having an impact on sites. We needed to have high availability and scalability and to be able to grow the processing power cheaply and easily. Why were some rejected?
Price point was huge issue [with EMC and Network Appliance]. We were in no way comfortable with the reliability and scalability of the Windows products. Nobody could give us any customer examples of people running a huge amount of storage behind a Windows box. Why did you choose Linux?
We chose Linux because of the scalability and cost. In a Linux cluster environment, all we have to do is get a new server and pop it in there. We have to have scalability like that because we've got 11 terabytes now, and we're planning on adding to that over the years. What daily hassles were caused by your aging, fragmented infrastructure?
Loading new content, backing up content, trouble-shooting problems across two different platforms became an administrative nightmare. What daily hassles were caused by your aging, fragmented infrastructure?
A lot of our customers use a site to research images to use in their projects, and that researching takes place via the site's search engine. The search engine led to the database, the database was on the same server as all of the content, and so the load on the search engine was too great. That slowed everything down, including the purchasing of [products].

This really started to become a problem within like the last year. We juggled around some stuff, and our DBA guys worked in the database to try to tune it a little bit better, but there weren't really any hardware upgrades we could do with what we had. What benefits would be gained from consolidating the sites?
A lot of the content that involves both of the sites is shared content. For example, you may go to PictureQuest.com and purchase an image, but you could also go to Creatas.com and purchase the same image. Going forward, there will be more content, not just photography but footage and bounds and bonds that will be shared across both of the sites. Having the content in two places becomes challenging; more disk space is used because the same exact files from two places have to be backed up, loaded and inventoried. What challenges did you face in changing your storage infrastructure?
We had to consolidate multiple e-commerce Web sites onto a single storage backbone serviced by a shared data cluster. We had to migrate from legacy Unix and Windows platforms to Linux. We had to improve availability and performance. Could you give us a picture of Dynamic Graphics' IT world prior to the Linux migration?
We had a variety of Unix and Windows servers along with NAS and SAN storage technologies. For our two main e-commerce sites, we had PictureQuest on Unix (Solaris) and Creatas.com on Windows. Why was a change needed?
The driving factor for the change in our infrastructure was the need to consolidate all of the Web sites onto a single platform for manageability and scalability, as well as to have a single source of content that is shared across the different sites. We did not have the hardware in place to grow either one of those platforms separately or to be able to accommodate the sum of both of the sites together. Also, neither one of the platforms as they stood had the ability to scale to accommodate having all that content in one place.

In a Linux cluster environment, all we have to do is get a new server and pop it in there. We have to have scalability like that because we've got 11TB now, and we're planning on adding to that.
Brian Kohles
Web developerDynamic Graphics Group
Why didn't you just update your legacy systems?
We had Windows 2000 running app servers, and our SAN and database was on Windows NT 4.0. That software was getting dated. Our Unix hardware was old, too. Our SAN was a first-generation Compaq SAN, and the drives were no longer available. We needed to increase drive size, and the cabinet was full.

Dig deeper on Linux high-performance computing and supercomputing

0 comments

Oldest 

Forgot Password?

No problem! Submit your e-mail address below. We'll send you an email containing your password.

Your password has been sent to:

-ADS BY GOOGLE

SearchDataCenter

SearchServerVirtualization

SearchCloudComputing

SearchEnterpriseDesktop

Close