PostgreSQL can be used in a lot of different ways
The relational DBMS industry is filled with startups. In some way or other, most of them are based on or make use of the open source project PostgreSQL. (Not all, of course; exceptions include DATAllegro and Infobright, which are based on Ingres and MySQL respectively.) But how they use PostgreSQL varies greatly.
EnterpriseDB is at one extreme. It hired a number of the top PostgreSQL developers, and is widely credited in the industry for major enhancements in PostgreSQL 8.3. Look in the future for EnterpriseDB to be less “The Oracle-compatible database company” and more “The PostgreSQL company that offers great features, including Oracle compatibility.”
Vertica is at the other pole. I still have the impression that Vertica emulates PostgreSQL to gain various tool compatibility benefits. But upon checking I learned, to my surprise, that Vertica uses no actual PostgreSQL code whatsoever.
Greenplum is somewhere in between. Greenplum started from PostgreSQL, and said it was supporting something open source called Bizgres. But Seth Grimes calls Greenplum’s commitment to Bizgres into serious question. But then, given how silent Greenplum has been recently (some PR around an investment round excepted), I’m not sure what to think of anything I previously heard from the company.
Please subscribe to our feed!
Comments
10 Responses to “PostgreSQL can be used in a lot of different ways”
Leave a Reply
I think you should mention Netezza, which uses some of PostgreSQL code
While that’s true, it’s not to the same extent as EntDB or GP. NZ’s use is fairly limited, as far as I’m aware; it’s mostly used for the data dictionary and some tools, I believe.
Lots of outfits use PostgreSQL code to some extent. Hence my reference to “extremes.” Netezza and some of the CEP outfits are more in the middle.
CAM
Hello Curt. A number of points —
– Netezza’s DBMS is a PostgreSQL fork.
– DATAllegro, to be precise, is an appliance rather than a DBMS, and Ingres is the DBMS on the appliance. Similarly, for precision’s sake: Infobright has created an engine that plugs into the MySQL framework, essentially an alternative to the MyISAM engine, which was formerly the best choice for analytical work.
– Greenplum appears to be doing quite well, with significant growth in paid customers for their Greenplum DB (formerly known as Bizgres MPP).
– EnterpriseDB has some intentions of rivaling Greenplum as an enhancement of PostgreSQL for data warehousing. They’d do this by integrating PostgreSQL MPP technology they acquired for ExtenDB last summer-fall.
– It of course would make sense for Vertica to “emulate” PostgreSQL given that Stonebraker was behind Postgres and is behind Vertica. Vertica is a commercialization of the open-source academic project C-Store, just as Illustra was a commercialization of mid-’90s Postgres.
I’m almost ready to put (figurative) pen to paper and write up the report on “open source (based) data warehousing” that I’ve been talking about for 5-6 months now.
Seth,
Look again at the DATAllegro technology.
And when you say “formerly known a Bizgres MPP”, are you talking about sales in conjunction with Sun, or something else?
Best,
CAM
Look again at the DATAllegro technology?
See http://datallegro.com/images/v3_node_large.jpg . Ingres is the DBMS on the DATAllegro appliance.
According to a Feb 2007 press release, “DATAllegro has used the Ingres open source database in its data warehouse appliance since 2004”: http://datallegro.com/pr/2_20_07_datallegro_ingres.asp .
Bizgres MPP is what Greenplum DB used to be called. That’s on all supported platforms: Linux and OSX in addition to Solaris. The company changed the name of their only commercial product to match the company’s own name. The company actually used to be known as Metapa and their product was originally called Metapa Cluster Database.
Seth,
Please look at resources such as http://www.dbms2.com/2006/09/20/myths-datallegro-ingres-open-source/ or, perhaps even more to the point, Stuart Frost’s second comment on http://www.dbms2.com/2006/07/03/datallegro%E2%80%99s-technical-strategy/
DATAllegro doesn’t use the Ingres optimizer. DATAllegro query execution leaves Ingres multiple times during a typical query (because the Ingres instances run on single nodes only). There is no Ingres subsystem to DATAllegro’s software of which one can correctly say “Oh, Ingres is doing all the DBMSy parts, and what’s around is something other than a DBMS.” Any definition of “DBMS” under which Ingres is “the” DBMS on the DATAllegro appliance is not one I would endorse, and I suspect I’m in the large majority in using the term that way.
If it doesn’t build and execute the query plans, it’s not a DBMS to me. Ingres doesn’t do all of that for DATAllegro, so it’s not the whole DBMS.
Is this now clearer?
Best,
CAM
It is a very interesting market, with all the new rdbms popping up as an alternative to the old giant’s oracle/db2/mssql/sybase…
Though not all base their code on postgres.
We have 100% our own code base here at Scimore.com, with some intriguing features. E.g. horizontal scale out.
Anyone have an exhaustive list of the new rdbms up comers?
Regards, Thomas
Seth,
I talked today with Andy Astor, CEO of EnterpriseDB, and he confirmed that your view of their plans is inaccurate. Or maybe it’s your view of Greenplum that in my opinion is inaccurate. It will be a looooong time before EnterpriseDB supports more than the occasional data warehouse with >10 terabytes of user data. By way of contrast, the sweet spot of the data warehouse appliance market now is 5-30 Tb, and headed up quickly by another factor of 5.
Some vendors already aren’t very competitive below 30-50 Tb or so.
CAM
[…] use MySQL, although I wouldn’t want to commit to exact details. By way of comparison, PostgreSQL-compatible Vertica actually uses no PostgreSQL code at all, whereas row-based Greenplum and Aster Data make heavy use of PostgreSQL code. DATAllegro (now […]