IBM’s “Shark” Eaten by EMC’s “BASS”

Event Summary

EMC Corporation has launched a new high-end Symmetrix 8000 family, its latest entry into the enterprise storage market. Maximum capacity is 19.2 terabytes (TB), the highest/largest on the market, and surpassing IBM's Enterprise Storage Server by approximately eight TB. The top-of-the-line model is the 8730, which can hold up to 384 disks. In addition to the 8730, EMC announced the five-terabyte 8430.

In addition to the big boost in capacity (more than double that of the Symmetrix 5000), EMC has doubled the internal bandwidth (a/k/s "bus speed"), from 700 MS/sec to a whopping 1.4 GB/sec, far surpassing IBM's 800 MB/sec.

EMC's most serious competition currently comes from IBM, but Jim Rothnie, EMC's senior vice president of product management, has stated "That will be the closest time when our competitor will catch up to were we are now. Our lead has been extended to three years."

As if falling behind in the capacity race was not bad enough, IBM recently announced that basic Fibre Channel connectivity and data replication software for ESS has been delayed until August or September.

Market Impact

EMC's announcement sets the bar high enough that IBM will be hard-pressed to match it. Although IBM's ESS (The Product Formerly Known as Prince - er, uh, "Shark") was in the same general ballpark for enterprise storage - in fact its rated capacity exceeded EMC's former maximum - that proximity lasted only nine months.

The Symmetrix 8730 now has a 70% edge in max capacity, and a 75% edge in internal bandwidth/bus "speed" (1.4 GB/sec vs. 800 MB/sec). Although the performance/spec differential would be tough to sell against under optimal conditions, the Shark has been plagued by reports of glitches/bugs. It is unclear if IBM has resolved all the glitch issues, but their mere existence is a problem.

One other issue is the one-year gap between the Shark announcement and the expected release of the "Virtual Manager". Some feel the Virtual Manager is critical to ESS's ability to compete with EMC; making customers wait until late 2000 is probably not the best tactical maneuver. Add to that EMC's reputation for shipping high-quality - albeit expensive - systems, and you start looking at EMC increasing its share of the enterprise market (currently estimated at 35%).

EMC's senior management is predicting that they now have a three-year lead on IBM - in other words, IBM will not catch up until 2003. Although we admire EMC's self-confidence, we think such predictions may be premature. We don't necessarily disagree, since IBM is not known for its ability turn things around quickly, no matter how good they are at technology/research.

One downside to the new 8000s: they are not backward compatible, so they are not really an upgrade path. Rather, as EMC affirms, the 8000 series is for "new" storage. Of course, companies that can afford the 8730 may have sufficiently large IT budgets to be able to ignore the downside. It may not be a J.P. Morgan-esque scenario ("If you have to ask, you can't afford it"), but you get the idea.

User Recommendations

These are high-end systems - not for the small shop nor faint of heart. This announcement will affect (for now) a relatively small percentage of companies - those who need new storage capacity of terabytes. The vast majority of companies have built their data centers (or similar enterprise storage system) over time, adding on as needed.

For those who need the capacity, though, EMC certainly is worth investigating seriously. EMC's "quality" sales message carries a lot of weight; especially for those customers who realize that the high-upfront price redeems itself the first time the low-priced system dies (or corrupts data).

IBM's specs on their ESS are also good. Until April 19th, their specs were really good. The ESS should still be a contender, especially if IBM demonstrates glitch freedom to a customer's satisfaction. IBM's pricing (roughly equivalent to 25 cents per MB) is attractive, as well.

Customers should ask for multiple reference sites for each product, and look at existing customers whose IT infrastructure and strategy most match their own.