To: Mighty Mizzou who wrote (26268 ) 6/11/1999 3:59:00 PM From: Mighty Mizzou Read Replies (2) | Respond to of 77400
Big Blue picking on little ole Cisco! Waaaaaaaaahhhhhhhhhhh!!!!!!!! It's Not Fair!!! WWWWWAAAAAAAAAAHHHHHHHHHHHH!!!!!!!!!!!!! Test results spark IBM-Cisco battle IBM claiming Network Utility and 2216 support more tn3270 sessions for less than Cisco CIP. By MARC SONGINI Network World, 05/31/99 RESEARCH TRIANGLE PARK, N.C. - Using recent router-performance test results, IBM is delivering jabs at archrival Cisco, but Cisco is crying foul. IBM has revealed to Network World the results of a series of tests comparing the performance of the IBM 2216 multiaccess connector and the Network Utility tn3270 Server device to that of a Cisco 7507 router equipped with a Channel Interface Processor (CIP). All three devices were attached to an IBM G-5 mainframe and ran tn3270 sessions. According to tests conducted by The Tolly Group, an independent testing firm, IBM's Network Utility and 2216 outperformed the 7507 with the CIP by as much as 40%. The 2216 is a router that handles SNA and IP; the Network Utility is a smaller SNA-to-IP gateway; and the CIP is a card that sits in a Cisco IP router and links IP or SNAnets to a mainframe. Tn3270 software lets users access SNA resources via IP links. These SNA-IP gateway products are crucial for users who need to access their legacy applications over IP. Because the SNA applications are so important and accessed so frequently, an SNA-IP gateway can frequently be a source of bottlenecks. And the results are . . . IBM claims it has scored big in the latest tests. For instance, in a 9,000-session tn3270 test, the IBM 2216 handled 1,052 transaction/sec as compared with Cisco's 704 transaction/sec (TPS) - a 33% gap. By dividing the list price of the hardware and software by the number of transactions per second, these results translate into about $67 per TPS for the 2216, compared with $163 per transaction for Cisco's 7507. IBM claims its gear maintained this pricing difference over the majority of the tests. However, Frank Maly, head of marketing for Cisco's InterWorks Business Unit, says this test is just creating confusion for users. The test only focused on the performance of a single Cisco CIP - while IBM was using two full router platforms - the 2216 and Network Utility. Such a comparison is inaccurate, Maly says, charging that "it tells the customer nothing." However, in each test case, The Tolly Group tested a single gateway in isolation with a single channel attachment to ensure an even comparison among all of the systems. IBM offers two alternatives for customers and requested that both be tested. This isn't the first time IBM and Cisco have argued over gateway benchmarks. Originally, IBM held a 2216-75XX test in August 1997 (NW, Aug. 18, 1997, page 78), also conducted by The Tolly Group. The findings of those tests were favorable to IBM and disputed by Cisco. Since then, there have been several unsuccessful attempts to agree on testing terms acceptable to both companies. The two companies still couldn't work out their disagreements in time for this recent round of tests, held during March and April at an IBM facility in Gaithersburg, Md. Although Cisco's Maly originally committed to the project, he declined to participate before testing began. Maly says he refused because IBM wouldn't allow Cisco to fund half of the test, or co-design it: "We recognized this was an uneven playing field, and we chose not to participate." Maly says he has a standing offer to hold a Cisco-IBM bakeoff - if it is conducted fairly. IBM is touting the recent test results as a victory for itself in the price/ performance arena - its Network Utility costs about $62,500 and its 2216 is priced at $65,245. Cisco's 7507 with CIP costs $119,650. "We clearly have an edge in performance," says Jim Goethals, product manager at IBM's Networking Hardware Division. User leery of results But such IBM-sponsored tests are not to be trusted, says one Cisco user, an IT manager at a large insurance company who spoke on condition of anonymity. The user's network has 17 mainframes running IP and SNA to remote users and sites for a variety of business transactions. The previous Tolly test left the user unimpressed: "It was configured in a way that made IBM look a lot better than Cisco." In this user's network, there are 10 Cisco 75XX routers with 14 CIP cards running SNA and IP traffic. The user's company went with Cisco because the firm offered ATM LAN connectivity, which the IBM 2216 didn't have at the time, the user says. The Tolly Group, which is publishing the results on its Web page (www.tolly.com), insists its tests are objective and that the results can be repeated.