Two weeks in Honolulu: No, not for R&R, but to conduct InfoWorld's first SAN integration challenge. The first week was dedicated to testing three mid-range storage area networks, and the second week was set aside to integrate them.
Four vendors were invited to participate in the integration: EMC, Fujitsu Softek, Hewlett-Packard and IBM. The facility was the Advanced Network Computing Lab at the University of Hawaii, overseen by Lab Director Brian Chee.
The first week's test, managed by InfoWorld's Mario Apicella, brought together an EMC Clariion CX400 array, an HP StorageWorks EVA 3000, and IBM's TotalStorage SAN Integration Server. Included in the setup were Brocade Fibre Channel switches provided by each vendor, as well as a variety of storage disk arrays, servers, and client devices. IBM also supplied a Sun server to use for interoperability testing.
Moving into the second week's testing, the vendors (with help from Apicella and Chee) merged the three Fibre Channel networks into a single large fabric. This in itself was a challenge, but fortunately was achievable only because all three SANs happened to use the compatible Brocade switches.
That hurdle cleared, each of the four integration vendors were given one day to simulate a customer engagement, and tasked with integrating three SANs as tightly as possible. Each vendor team was handed a set of sample requirements for an integrated SAN, ranging from easy tasks like consolidating consoles, reporting, and alerting, all the way through creating pools of storage that would span each of the SANs. The company's engineers would have most of the day to install and configure their solution, followed by a presentation and scoring.
Truth be told, the integration test proved not completely representative of a real-world customer engagement, and thus, not entirely successful. The lack of prior knowledge of the hardware to be installed in the lab, the use of newly released equipment for which competitive drivers and agents were not available, the inability to set the network back to a known "natural" state between tests, and the fact that hardware vendors were wary of giving their competition unfettered access to their hardware, led to unexpected complications and compromises. No vendor, not even the winner, Fujitsu Softek, was fully able to meet all the requirements in the time allotted.
Because so many of the difficulties the vendors experienced were, in our opinion, a result of the test environment rather than indicative of the features and functionality of their specific solutions, we have dispensed with scorecards for this review, substituting instead a rough check list of task categories.
We've learned from this first effort: Future InfoWorld SAN integration tests will maintain much tighter control of the test hardware and conditions, as well as longer time slots allocated for each vendor. Despite those challenges, we believe the SAN integration test was a success, pointing to virtualization as the optimal solution, while also revealing the continuing complexity of storage area networking.