Monday, April 5, 2010

The new archive project started as a replacement for an existing disk archive system that they had purchased from a major storage supplier. The former system was designed to archive email to meet compliance concerns that the organization had because of their pharmaceutical research business. From the beginning, the original system had problems, starting with installation, which took three times as long for the vendor to complete than was projected. Also, the disk archive required a proprietary API set for the email archive software to work with it. There were problems with the API integration module from the software supplier and there were problems with the API set itself. This also made the device feel single-purpose in its deployment. Beyond that, the system was slow to respond, difficult to back up and very expensive to upgrade. The total costs were nowhere near the cost of tape storage, which was promised. There were also issues with general reliability of the hardware itself. An outage of two weeks was the final straw and triggered a search for a replacement.

Learning from their past mistakes the company put together a very extensive list of requirements for the perspective new supplier to meet, beyond what they considered 'disk archive basics'. First, it had to have an open protocol that was easy for a broad range of applications to support. Second, it had to perform well and in a scalable fashion by leveraging multiple network I/O paths to the system. Third, it had to do so in a much more cost effective manner than the previous system. The company really did want the system to approach the cost of tape.

After interviewing and extensive testing of several disk archive vendors as well as considering cheap NAS, the company selected Permabit as their archive platform. It met all the base criteria the company expected in their original archive including, scalability, retention and reliability, but also did so with excellent performance and at a very aggressive price point which would help them approach the cost of tape consideration mentioned previously. In fact, they were able to obtain the Permabit on-site archive and a second archive which would be a replication target at their DR site, for the cost of the original single site email archive.

The initial project was to use the Permabit system to replace the current disk archive email solution; it was up and ready to receive data in a few days. Integration of the email archive application to the disk archive was as simple as pointing the software at a network mount point. Permabit uses standard CIFS and NFS data shares to present the archive to applications which opened up a wide variety of uses for the company to leverage the archive investment. Finally, users saw a noticeable increase in overall system performance when retrieving old emails.

As the new system was coming into production, a new requirement was arising elsewhere in the data center. The finance group needed the ability to permanently store copies of expense reporting data and do so in a fashion where that data would be unalterable. The initial thought, prior to the Permabit system installation, was as this data came in, to store this data to a network file server first, then when enough information had queued up move it to a WORM CD. Ideally the goal was to make sure that the data was never stored by the company in an alterable format and storing on a standard file server share first would not meet this goal.

With the email archive project moving ahead of schedule, the storage manager proposed another use for the Permabit archive system. Partly this was because of the smooth installation and the pace at which that project was moving, but also because of an unexpected benefit of the archive system; capacity optimization. While the original system would eliminate redundant files, known as single instancing, the new archive system would identify redundancy at a sub-file level, known as deduplication. As a result, redundancy between similar files, a common occurrence in their email stores, was eliminated. In addition, the new system would also compress data prior to deduplicating it, further increasing the space savings. The net result was that the project was ahead of schedule and under budget, both from a cost and capacity standpoint.

Integration of this new project, the storage of expense data, was almost instant. Again, leveraging Permabit's standard protocol presentation, the storage manager simply created a Windows share for the finance team and they wrote the expense data directly to the system. The direct write, which was enabled by the enhanced performance of the new system, coupled with the built-in WORM capability solved the problem of making sure the data was only stored in an unalterable fashion.

The success of this additional project triggered several other use cases and lead to a need for additional storage capacity. One of the key decision points was the affordability of additional capacity. What the company has found is that the Permabit expansion is extremely cost effective, because Permabit stays in line with current drive prices and passes the cost reduction onto their client base rather than keep it as margin as many of the larger vendors do. They have also found each upgrade to be seamless and non-disruptive to the process.

This combined success lead the pharmaceutical company to use the Permabit system as a backup and archive repository for their VMware images. Almost everything in a pharmaceutical company needs to be retained and VMware images are no different. While the performance of the system had more than met expectations, the storage manager knew that it was not designed to be a backup target.

Fortunately, his selection of Vizioncore vRanger worked perfectly with the Permabit Archive because of its ability to do block level incremental backups of the server images, resulting in only a fraction of the actual data being sent to the archive at any given point in time. The process leverages these incremental backups plus the archive's deduplication capabilities to be able to back up thousands of copies of VMware images and hundreds of versions of each of those images as they are changed. The company also leveraged Permabit's replication capability to automatically off-site server images to their DR location. The net result was a complete set of server images ready to be deployed at a second site in the case of a disaster.

By selecting an open disk archive the pharmaceutical company was able to expand the use of the solution across many areas within the organization. While a rapid ROI was achieved on the email archive project alone, the ROI was repeated many times over because of Permabit Enterprise Archive's capabilities in open architecture, scalability and cost effectiveness.

George Crump, Senior Analyst