Re: Informix vs. Sybase vs. Oracle vs. (gasp) MS SQL Server

From: Perry Dillard <perry.dillard_at_iname.com>
Date: 1997/11/24
Message-ID: <34799917.5002172_at_surge.sfs.com>#1/1


Obviously Pablo has never written an OLTP Order Entry system where multiple users are constantly updating an inventory table and committing quantites to customers on the fly. If he had he would know that row level locking is absolutely essential and that page level will not let you see the realtime up to the instant available quantities for a given inventory item.

I'll admit that page level locking has its place, but row level is absolutely essential. And I don't believe anyone has tried to argue that row level locking carries no overhead. Of course it has more overhead than page locking. But it mostly depends on how many rows are actually being locked. If you keep one row locked in a header table during an update transaction, you aren't incurring a significant amount of overhead compared to a page lock. And you get the added benefit that someone else can be modifying the record right next to you since you don't have a whole page of data locked. This is quite critical when you have a multi-user application.

-perry

On 21 Nov 1997 07:00:32 -0700, Pablo Sanchez <pablo_at_sgi.com> wrote:

>[ this message is so old ... ]
>
>>>>>> "Chris" == Chris Jenkins <chrisj_at_informix.com> writes:
>Chris>
>Chris>
>Chris> Sorry but it is a *big* limitation. Row level locking provides much greaterconcurrency,
>Chris> even for a well designed application database.
>Chris>
>
>Think again. Look at the current -C values for Sybase and
>look at what Informix is generating. Sounds like you fell
>for that one too...
>
>Chris> The only reason that the default is page level locking is historical. It usedto be that
>Chris> way so we haven't changed the default to maintain backwards
>Chris> compatibility. We always encourage use of row level locking these days.
>Chris>
>
>I don't think so... common sense dictates that there be
>overhead via your lock manager. More locks, more overhead.
>Anyway, this is a futile discussion because it's been hashed
>out over and over and over and over again...
>--
>Pablo Sanchez | Ph # (650) 933.3812 Fax # (650) 933.2821
>pablo_at_sgi.com | Pg # (800) 930.5635 -or- pablo_p_at_pager.sgi.com
>-------------------------------------------------------------------------------
>I am accountable for my actions. http://reality.sgi.com/pablo [ /Sybase_FAQ ]
Received on Mon Nov 24 1997 - 00:00:00 CET

Original text of this message