Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Technology Stocks : IFMX - Investment Discussion -- Ignore unavailable to you. Want to Upgrade?

To: James A. Shankland who wrote (9811)	3/5/1998 1:07:00 AM
From: harryl1	Read Replies (1) \| Respond to of 14631

I had some folks return from LA where MSFT was unveiling the latest SQL Server 7.0 beta two a.k.a. sphinx. I thought IFMX on WinNT was MUCH better than SQL Server 6.5 -- but -- with infinite money, MSFT has basically re-written most all of the piece of crap they called SQL Server and built something that does scare me a little -- now that they have the basics right -- or will, by 7.x, the "only" challange left is scalability (which merced, clustering, etc will eventually nail). I still think we have two good years at the high end without thinking about MSFT, but it is already scary how much big stuff is done with WinNT clusters, and their seat pricing will create some downward pressure (But not the $100 per seat vs. $2000 per seat we once feared.) I will print highlights of sql server update in next message for those interested...

To: James A. Shankland who wrote (9811)	3/5/1998 1:12:00 AM
From: harryl1	Read Replies (1) \| Respond to of 14631

OT not really, for you long term investors -- here is some impressions of sql server 7.0: The following note is a "dump" of the notable things Carlos & I heard/saw at the Microsoft SQL Server 7.0 (a.k.a. Sphinx) Workshop held in Los Angeles 2/25-2/27/98. I classified them as notable because I feel they either will change or could change our "SQL philosophies" on XXXXXXX (our product). As you can see, there are a truck load of changes with Sphinx and therefore I've given the highlights very tersely. I will gladly expand upon any item for you individually or, if there is sufficient interest, in a meeting for the group. New in SQL Server 7.0 (no particular order) - row level locking - data communications with the server exclusively Unicode - ISQL changed from DB-lib based to ODBC based - OLE DB consumes/produces native TDS (Sybase's tabular data stream protocol) for high performance - the 1mb query limit has been removed; in 6.x sort and hash ops were cramped by this constraint - datatypes are no longer converted to strings for transmission... the native datatypes are sent - Microsoft Transaction Server's connection pooling feature can reduce quantity of expensive connections - TEXT datatype is improved to be more like "a long character" type - new datatypes: GUID type (universally unique a replacement for token mgr!, facilitate parallel PowerFAIDS operations), Unicode datatypes, - various limits have been removed (e.g. number of cols in a table, number of tables in a join, max char length, etc..) - physical structure of indexes - the relationship between Cluster Indexes and Secondary Indexes; want narrow clustering indexes since secondary indexes are referenced through them; don't want long strings in clustering index -- disk hog - CONCAT_NULL_YIELDS_NULL - MS currently developing a JDBC driver for 7.0 - JDBC is cross platform strategy rather than writing OLE DB data providers for other platforms e.g. Sun - storage engine performance improvements (e.g. asynch i/o, optimistic reads, - Query Processor/ Transact SQL performance improvements (e.g. compiled, shared query plans, plans are shared between users!, plans for views, SPs, even ad hoc queries! shared plans cut down on system table contention, - DMO (Distributed Management Objects) for administrative tasks (e.g. user.add) ALSO use DMO for Web-based remote server administration (e.g. XXXXXX CSRs use it rather than FTPing databases). - our current "Connection Busy...HSTMT..." error has not been fixed in 7.0 and is not in the plan; it is a wart from the Sybase half duplex TDS protocol - Microsoft Management Console -- MS's leapfrog over PowerBuilder's Database Painter - Profiler -- analyses queries and runs them through the real QP to determine plan - Record/Playback of SQL workload -- allows debugging, bug replication (e.g. QA tool), performance tuning (add one index w/o breaking performance somewhere else), - major rework of SS positions MS to "have more rapid releases... about 1/year" - ODBC 3.x (3.7 appeared on one slide) default behavior is 2.x; you must explicitly turn on 3.x - ODBC 3.x is mainly Unicode support; does include error handling improvements to SQLError - column level DDL (add/modify/drop) - auto dealloc cursor for Transact-SQL - multiple triggers per trigger type (e.g. more than one insert trigger on stu_award). - user defined functions are a post 7.0 priority - user defined datatypes are a post 7.0 priority - should declare RI even though we need triggers to implement cascade; optimizer uses the RI keys to prune the plan - debugging of SPs and triggers available in VC++, VB and VInterdev - parallel query execution on SMP boxes (not clusters) -- clusters post 7.0 priority - "Auto Management" e.g. disk growth & SHRINKING!; policy issues remain with shrinking and will be changed in Beta III - physical storage no longer "device based", now file group and file based - current statistics are essential -- leave auto stats on - desktop versions of SS 7.0 (e.g. Win 9x) same code base as Server based versions; small footprint