>>1) There are several instructive white papers on REDB's web site
>>that nicely explain data warehousing and how REDB is different.
If you look at the white paper dates, only one on data warehouse architecture is since Sep 96. Meanwhile, the leaders have each made major changes since then to really change the field.
First, Sybase IQ with its unique vertical storage together with its bitmap storage/indexing can provide very fast responses on low cardinality data, especially when only a few fields need to be examined to determine the result set.
Second, IFMX (and Oracle) have implemented multi-table join indices, which is the same as or equivalent to the star-schema of Red Brick. IFMX details this in a white paper about its DSS indices (which also include the bit-map indices) that are part of its XPS 8.2 release.
Third, load and cleansing processes have been dramatically improved in the competitor DBMS's since those white papers were written. In particular, IFMX is capable of loading and deleting (including index adjustments) while other queries are running. This is part of the TPC/D multi-user benchmark, for example.
In summary, a lot of the problems of Red Brick are because its competitors have caught up to it in the areas of its strengths, and have invaded its niche. Further, they have shown much greater ability to scale. For example, the MCI warehouse on IFMX is over 1T of raw data, and this has now been in production for about 2 years. This seems to be far beyond the capacity of Red Brick. |
